Confidential...
Transcript of Confidential...
CSTATUS OF THESIS
Title of thesis
I, ABRAHAM DEBEBE WOLDEYOHANNES, hereby allow my thesis to be placed at the
Information Resource Center (IRC) of Universiti Teknologi PETRONAS (UTP) with the
following conditions:
1. The thesis becomes the property of UTP
2. The IRC of UTP may make copies of the thesis for academic purpose only
3. The thesis is classified as
If the thesis is confidential, please state the reason: _________________________________________________________________________________________________________________________________________________________________________________________________________________________
The contents of the thesis will remain confidential for ________years.
Remarks on disclosure: _____________________________________________________________________________________________________________________________________________________________________________________________________________________________
Endorsed by
_______________________ ______________________
Signature of Author Signature of Supervisor
Date:_______________________ Date:____________________
Confidential
√ Non-Confidential
Permanent Address: Addis Ababa University, Faculty of Technology, Addis Ababa, Ethiopia
Simulation Model for Pipeline Network System with Non-pipe
Elements for Natural Gas Transmission
AP. Ir. Dr. Mohd Amin Abd Majid
UNIVERSITI TEKNOLOGI PETRONAS
SIMULATION MODEL FOR PIPELINE NETWORK SYSTEM WITH
NON-PIPE ELEMENTS FOR NATURAL GAS TRANSMISSION
by
ABRAHAM DEBEBE WOLDEYOHANNES
The undersigned certify that they have read, and recommend to the Postgraduate Studies Programme for acceptance this thesis for the fulfilment of the requirements for the degree stated.
Signature: ____________________________________ Main Supervisor: Assoc. Prof. Ir. Dr. Mohd Amin Abd Majid
Signature: ______________________________________ Head of Department: Dr. Ahmad Majdi Bin Abdul Rani
Date: ______________________________________
SIMULATION MODEL FOR PIPELINE NETWORK SYSTEM WITH
NON-PIPE ELEMENTS FOR NATURAL GAS TRANSMISSION
by
ABRAHAM DEBEBE WOLDEYOHANNES
A Thesis
Submitted to the Postgraduate Studies Programme
as a Requirement for the Degree of
DOCTOR OF PHILOSOPHY
MECHANICAL ENGINEERING
UNIVERSITI TEKNOLOGI PETRONAS
BANDAR SERI ISKANDAR,
PERAK
MAY 2010
iv
DECLARATION OF THESIS
Title of thesis Simulation Model for Pipeline Network System with Non-pipe Elements
for Natural Gas Transmission
I , ABRAHAM DEBEBE WOLDEYOHANNES, hereby declare that the thesis is based on
my original work except for quotations and citations which have been duly acknowledged. I
also declare that it has not been previously or concurrently submitted for any other degree at
UTP or other institutions.
Witnessed by ________________________________ __________________________ Signature of Author Signature of Supervisor Permanent address: Addis Ababa University, AP. Ir. Dr. Mohd Amin Abd Majid Faculty of Technology, Addis Ababa, Ethiopia
Date : _____________________ Date : ________________
v
DEDICATION
To my beloved wife, Meaza Feyssa, and my daughter, Halleluiah Abraham,
for being with me during this time
vi
ACKNOWLEDGEMENTS
First and for most, thanks to my GOD for giving me the patience to start and
finalize this thesis work. I would like to extend my sincere appreciation to my
supervisor, Assoc. Prof. Ir. Dr. Mohd Amin Abd Majid, for his valuable support and
guidance throughout the completion of this thesis. He spent many of his precious
time to teach and guide me. His encouragement, creative suggestions, critical
comments and everlasting enthusiastic support have greatly contributed to this thesis.
I would also like to thank my beloved wife, Meaza Feyssa Tolla, for her love and
support during all the time. Not only has she encouraged me to pursue my academic
interests, she also sacrificed a lot to be with me during this time. For this, I will be
forever grateful and acknowledge with gratitude and love the importance of her in
my life. Furthermore, great thanks to my daughter Halleluiah for bringing me many
funs and pleasures in my life.
I would also like to thank the Universiti Teknologi PETRONAS (UTP) for
providing me grant and facility for the research. Furthermore, great thanks for Addis
Ababa University for granting my study leave to continue with my graduate studies. I
would also like to express my gratitude to PETRONAS GAS BERHAD for the
information and data about the gas pipeline networks. Great thanks to Dr. Dereje
Engida for revising the thesis and forwarding his comments. It is my pleasure to
acknowledge the support from Mr Noraz Al-Kahair Bin Noran for translating the
abstract to Bahasa Melayu.
Last but not least, great thanks to my fellow friends for providing me a good
atmosphere during my stay at UTP. There are also many people at UTP in one way
or another who have helped me. Thank you very much.
vii
ABSTRACT
The ever growing demand for natural gas enhances the development of complex
transmission pipeline network system (TPNS) which requires simulation procedures
for design and operation of the network. TPNS simulation is usually performed in
order to determine the nodal pressures, temperatures and flow variables under
various configurations. These variables are essential for analyzing the performance
of the TPNS. The addition of non-pipe elements like compressor stations, valves,
regulators and others make TPNS simulation analysis more difficult.
A new simulation model was developed based on performance characteristic of
compressors and the principles of conservation of energy and mass of the system to
analyze TPNS with non-pipe elements for various configurations. The TPNS
simulation model analyzes single phase gas flow and two-phase gas-liquid flow. The
simulation model also takes into account temperature variation and age of the pipes.
The model was designed for the evaluation of the unknown pressure, flow and
temperature variables for the given pipeline network. The two solution schemes
developed were iterative successive substitution scheme for simple network
configurations and a generalized solution scheme which adopt Newton-Raphson
algorithm for complex network configurations. The generalized Newton-Raphson
based TPNS simulation model was tested based on the three most commonly found
network configurations, namely: gunbarrel, branched and looped pipeline. In all the
tests conducted, the solutions to the unknown variables were obtained with a wide
range of initial estimations. A maximum of ten iterations were required to get
solutions to nodal pressure, flow and temperature variables with relative percentage
errors of less than 10-11.
viii
The results of TPNS simulation model were compared with Newton loop-node
method based on looped pipeline network configurations and an exhaustive
optimization technique based on gunbarrel pipeline network configuration. For both
cases, the results indicate that the model is able to provide solutions similar to the
compared models. In addition, the TPNS simulation model provides detail
information for the compressor stations. This information is essential for evaluation
of the performance of the system.
The application of the TPNS simulation model for real pipeline network system
was also conducted based on existing pipeline network system. Three modules of
TPNS simulation model which included input parameter analysis, function
evaluation and network evaluation module were evaluated using the data taken from
the real system. Analyses of the performance of compressor for existing pipeline
network system which included discharge pressure, compression ratio and power
consumption were also conducted using the developed TPNS simulation model. The
performance characteristics maps generated by the developed TPNS simulation
model show the variation of discharge pressure, compression ratio, and power
consumption with flow rate similar to the one available in the literatures.
Based on the results from the simulation tests and validation of the model, it is
noted that the developed TPNS simulation model could be used for performance
analysis to assist in the design and operations of transmission pipeline network
systems.
ix
ABSTRAK
Permintaan yang semakin tinggi untuk gas asli memerlukan pembangunan
kompleks sistem penghantaran saluran paip berangkaian (transmission pipeline
network system; TPNS) ditingkatkan di mana ia memerlukan prosedur-prosedur
simulasi untuk rekaan dan operasi rangkaian. Simulasi TPNS biasanya digunakan
bagi tujuan untuk menentukan pelbagai tekanan, suhu dan pembolehubah aliran
berbuku di bawah pelbagai konfigurasi. Pembolehubah ini adalah penting untuk
mengkaji prestasi TPNS. Penambahan elemen selain paip seperti stesen pemampat,
injap, pengatur dan lain-lain menyebabkan analisis simulasi TPNS lebih sukar.
Model simulasi baru dimajukan berasaskan kepada prestasi mesin pemampat dan
prinsip penjimatan tenaga dan juga jisim pada sistem bagi menganalisis TPNS
dengan unsur selain paip untuk pelbagai konfigurasi. Model simulasi TPNS
menganalisis aliran gas satu-fasa dan aliran cecair gas dua-fasa. Model simulasi ini
turut mengambil kira pelbagai suhu dan usia paip itu.
Model ini direka untuk menilai pembolehubah tekanan, aliran dan suhu yang
tidak diketahui untuk rangkaian saluran paip. Dua skim penyelesaian yang dirangka
adalah skim lelaran penggantian berturut-turut untuk konfigurasi rangkaian mudah
dan skim penyelesaian am yang menggunakan algoritma Newton-Raphson untuk
konfigurasi rangkaian yang sukar. Model simulasi am Newton-Raphson berasaskan
TPNS diuji berdasarkan tiga rangkaian paling umum yang didapati di konfigurasi
rangkaian iaitu: laras, bercabang dan lingkaran paip. Dalam semua ujian yang
dijalankan, penyelesaian pada pembolehubah yang tidak dikenali itu diperolehi
dengan satu anggaran julat awal yang luas. Iterasi yang diperlukan adalah sepuluh
lelaran untuk mendapatkan penyelesaian untuk pembolehubah tekanan, aliran dan
suhu berbuku dengan peratusan kesilapan relatif kurang daripada 10-11.
Keputusan model simulasi TPNS dibandingkan dengan kaedah gelung nod
Newton berdasarkan konfigurasi menggelung rangkaian saluran paip dan teknik
pengoptimuman yang lengkap berdasarkan konfigurasi laras rangkaian saluran paip.
Untuk kedua-dua kes, keputusan menunjukkan bahawa model ini boleh menyediakan
x
penyelesaian yang mirip dengan model yang sebanding dengannya. Seperkara lagi,
model simulasi TPNS menyediakan maklumat terperinci untuk stesen pemampat.
Maklumat ini adalah penting untuk penilaian prestasi sistem.
Penggunaan model simulasi TPNS juga diterapkan untuk sistem rangkaian
saluran paip sebenar berdasarkan saluran paip sistem rangkaian yang sedia ada. Tiga
modul bagi model simulasi TPNS ini termasuklah analisis parameter input, penilaian
fungsi dan penilaian rangkaian modul yang dinilai menggunakan data yang diambil
dari sistem sebenar. Analisis prestasi pemampat untuk sistem rangkaian saluran paip
sedia ada termasuklah tekanan luahan, nisbah mampatan dan penggunaan kuasa juga
dikendalikan dengan menggunakan model simulasi TPNS yang dibangun kan. Ciri
prestasi ‘peta’ dihasilkan oleh model simulasi TPNS yang dibangun kan
menunjukkan pelbagai tekanan luahan, nisbah mampatan, dan penggunaan kuasa
dengan kadar aliran yang menyerupai dengan hasil kajian orang lain.
Berdasarkan keputusan dari simulasi dan pengesahan model ini, ia menunjukkan
model simulasi TPNS yang dibangun kan boleh digunakan untuk menganalisis
prestasi bagi memudahkan reka bentuk dan pengendalian sistem penghantaran
saluran paip berangkaian.
xi
COPYRIGHT PAGE
In compliance with the terms of the Copyright Act 1987 and the IP Policy of the university, the copyright of this thesis has been reassigned by the author to the legal entity of the university,
Institute of Technology PETRONAS Sdn Bhd.
Due acknowledgement shall always be made of the use of any material contained in, or derived from, this thesis.
© Abraham Debebe Woldeyohannes, 2010
Institute of Technology PETRONAS Sdn Bhd All rights reserved.
xii
TABLE OF CONTENTS
DECLARATION OF THESIS ................................................................................. iv
DEDICATION ........................................................................................................... v
ACKNOWLEDGEMENTS ..................................................................................... vi
ABSTRACT ............................................................................................................ vii
ABSTRAK ............................................................................................................... ix
COPYRIGHT PAGE ................................................................................................ xi
TABLE OF CONTENTS ........................................................................................ xii
LIST OF TABLES ................................................................................................. xvi
LIST OF FIGURES .............................................................................................. xviii
NOMENCLATURES ........................................................................................... xxiii
PREFACE ........................................................................................................... xxvii
Chapter
1. INTRODUCTION .................................................................................................. 1
1.1 Background .................................................................................................. 1
1.2 Problem Statement ....................................................................................... 4
1.3 Research Objective ...................................................................................... 6
1.4 Research Scopes .......................................................................................... 7
1.5 Structure of Thesis ....................................................................................... 8
2. LITERATURE REVIEW .................................................................................... 10
2.1 Introduction ............................................................................................... 10
2.2 Optimization of Gas Pipeline Networks .................................................... 11
xiii
2.2.1 Dynamic Programming ...................................................................... 12
2.2.2 Mathematical Programming ............................................................... 15
2.2.3 Expert System .................................................................................... 17
2.2.4 Analytical Hierarchical and Network Reduction ............................... 19
2.3 Simulation of Gas Pipeline Networks ....................................................... 21
2.3.1 Simulation of Gas Networks without Non-pipe Elements ................. 22
2.3.2 Simulation of Gas Networks with Non-pipe Elements ...................... 26
2.4 Simulation Models with Two-phase, Temperature and Corrosion ........... 30
2.4.1 Two-phase Flow Models .................................................................... 31
2.4.2 Simulation Model with Temperature Variations ............................... 37
2.4.3 Internal Corrosion .............................................................................. 39
2.5 Solution Principles Applied in Gas Pipeline Network Simulations .......... 41
2.6 Summary.................................................................................................... 42
3. METHODOLOGY .............................................................................................. 45
3.1 Introduction ............................................................................................... 45
3.2 Mathematical Formulation of the Simulation Model ................................ 47
3.2.1 Single Phase Flow Equations Modeling ............................................ 47
3.2.2 The Looping Conditions .................................................................... 50
3.2.3 Compressor Characteristics Equations ............................................... 52
3.2.4 Mass Balance Equations .................................................................... 60
3.3 Successive Substitution based TPNS Simulation Model .......................... 61
3.3.1 Case 1A: Single Compressor Station and Two Customers Module .. 66
3.3.2 Case 2A: Two Compressor Stations and Two Customers Module .... 69
xiv
3.4 Newton-Raphson based TPNS Simulation Model .................................... 72
3.4.1 Case 1B: Gunbarrel Pipeline Network Module .................................. 76
3.4.2 Case 2B: Branched Pipeline Network Module .................................. 79
3.4.3 Case 3B: Looped Pipeline Network Module ...................................... 81
3.4.4 Case Study: Malaysia Gas Transmission System ............................. 85
3.5 Enhanced Newton-Raphson based TPNS Simulation Model .................... 88
3.5.1 Two-phase Gas-liquid Flow Modeling .............................................. 88
3.5.2 The Effect of Internal Corrosion ........................................................ 96
3.5.3 Modeling Temperature Variations ................................................... 101
3.6 Performance Evaluation of the TPNS Simulation Model ....................... 103
3.7 Summary .................................................................................................. 104
4. RESULTS AND DISCUSSIONS ..................................................................... 107
4.1 Introduction ............................................................................................. 107
4.2 Successive Substitution based TPNS Simulation Model ........................ 107
4.2.1 Case 1A: Single Compressor Station and Two Customers Module . 108
4.2.2 Case 2A: Two Compressor Stations and Two Customers Module .. 110
4.2.3 Limitations of Successive Substitution TPNS Simulation Model ... 112
4.3 Newton-Raphson based TPNS Simulation Model .................................. 114
4.3.1 Case 1B: Gunbarrel Pipeline Network Module ................................ 115
4.3.2 Case 2B: Branched Pipeline Network Module ................................ 120
4.3.3 Case 3B: Looped Pipeline Network Module .................................... 123
4.3.4 The Malaysia Gas Transmission Network Case Study .................... 132
4.3.5 The Case of Divergent in TPNS Simulation Model ......................... 143
xv
4.4 Enhanced Newton-Raphson based TPNS Simulation Model ................. 144
4.4.1 Two-phase Gas-liquid Flow Analysis .............................................. 144
4.4.2 The Effect of Internal Corrosion ...................................................... 150
4.4.3 Non-isothermal Flow Analysis ........................................................ 161
4.5 Verification and Validation of the TPNS Simulation Model .................. 163
4.5.1 TPNS Conceptual Model Validation and Verification .................... 165
4.5.2 Operational Validation ..................................................................... 167
4.6 Summary.................................................................................................. 177
5. CONCLUSIONS AND RECOMMENDATIONS ........................................... 179
5.1 Conclusions ............................................................................................. 179
5.2 Contributions of the Research ................................................................. 181
5.3 Recommendations ................................................................................... 183
REFERENCES ....................................................................................................... 185
APPENDICES
A. Snapshot of the Developed TPNS Simulation Code using Visual C++
B. Results of TPNS Simulation Model for Gunbarrel Network
C. Successive Substitution based Simulation Results (Case 1A)
D. Divergent Case in TPNS Simulation (Case 2C)
xvi
LIST OF TABLES
Table 2.1 Gas flow equations and range of applications adapted from [7] and [23]
................................................................................................................................ 25
Table 2.2 Summary of TPNS Simulation Models .................................................. 43
Table 3.1 Single phase flow equations and functional representations .................. 49
Table 3.2 Values of coefficients for compressor equations .................................... 56
Table 3.3 Number of equations and unknowns for TPNS configurations .............. 62
Table 3.4 Gas properties ......................................................................................... 63
Table 3.5 Summary of flow equations and their corresponding functional
representations for the given TPNS ........................................................................ 67
Table 3.6 Pressure data for single CS and two customers ...................................... 68
Table 3.7 Pipe data for single CS and two customers ............................................ 68
Table 3.8. Summary of flow equations and their corresponding functional
representations for the given TPNS ........................................................................ 70
Table 3.9 Summary of compressor equations and functional representations ....... 70
Table 3.10 Pipe data for two CSs and two customers ............................................ 72
Table 3.11 Summary of flow equations for the given gunbarrel TPNS ................. 77
Table 3.12 Summary of compressor equations and functional representations ..... 78
Table 3.13 Pressure requirements for two CSs and single customer ...................... 78
Table 3.14 Pipe data for two CSs and single customer .......................................... 78
Table 3.15 Summary of flow equations and their functional representation .......... 80
Table 3.16 Summary of mass balance equations and functional representations ... 80
Table 3.17 Pipe data for single CS and five customers .......................................... 81
xvii
Table 3.18 Summary of flow equations and functional representations ................ 83
Table 3.19 Summary of mass balance equations and functional representations .. 84
Table 3.20 Pipe data for single CS and eight customers ........................................ 85
Table 3.21 Two-phase flow equations and functional representations .................. 93
Table 3.22 Summary of flow equations for the given gunbarrel TPNS ................. 94
Table 3.23 Summary of flow equations and their functional representation ......... 95
Table 3.24 Summary of flow equations for the given gunbarrel TPNS ................. 99
Table 3.25 Summary of flow equations and their functional representation
considering corrosion ........................................................................................... 100
Table 3.26 Summary of temperature equations for gunbarrel TPNS ................... 103
Table 3.27 Various cases considered for the analysis using developed TPNS
simulation model .................................................................................................. 106
Table 4.1 Results of nodal pressures and flow variables (case 1B) ..................... 115
Table 4.2 Results of nodal pressures [47] ............................................................ 118
Table 4.3 Results of nodal pressures after ten iterations ...................................... 124
Table 4.4 Results of main and branch flow variables after 10 iterations ............. 125
Table 4.5 Pipe data [7] .......................................................................................... 128
Table 4.6 Pressure and flow requirements [7] ...................................................... 129
Table 4.7 Results of input analysis for pipes ........................................................ 134
Table 4.8 Results of nodal pressures and flow variables after 10 iterations ........ 136
Table 4.9 Comparison of nodal pressures for different pipe ages ........................ 153
Table 4.10 Variation of flow with ages of pipes .................................................. 160
Table 4.11 Results of temperature variables for the first 10 iterations ................ 161
xviii
LIST OF FIGURES
Figure 2.1 Gunbarel structure transmission pipeline network adapted from [8] .... 13
Figure 2.2 Flow of the gas though pipe adapted from [7] ...................................... 24
Figure 2.3 Typical performance map of centrifugal compressor [59] .................... 29
Figure 2.4 Flow pattern in horizontal and near-horizontal pipes [70] .................... 34
Figure 2.5 Schematic of homogeneous no-slip flow model adapted from [70] ..... 36
Figure 2.6 Analysis of temperature variations adapted from [23] .......................... 38
Figure 2.7 Effect of years in service on pipe roughness [22] ................................ 40
Figure 3.1 Structure of the simulation model for transmission pipeline networks . 46
Figure 3.2 Pipe joining two consecutive nodes ...................................................... 48
Figure 3.3 Part of transmission pipeline network system ....................................... 49
Figure 3.4 Looped pipeline network system........................................................... 51
Figure 3.5 Normalized head and normalized flow rate for different speeds of a
typical centrifugal compressor ................................................................................ 53
Figure 3.6 Efficiency and normalized flow rate for different speeds of a typical
centrifugal compressor............................................................................................ 54
Figure 3.7 Compressors operating in series within a station .................................. 56
Figure 3.8 Compressors working in parallel within a station ................................ 58
Figure 3.9 Comparison of selected data and approximated data based on the
compressor data from [59] ...................................................................................... 59
Figure 3.10 Comparison of calculated and actual pressure ratios ......................... 60
Figure 3.11 Mass balance formulation ................................................................... 61
Figure 3.12 Possible information flow blocks for compressor .............................. 64
xix
Figure 3.13 The developed flow chart for successive substitution based TPNS
simulation model .................................................................................................... 65
Figure 3.14 Gas transmission system for two customers ....................................... 66
Figure 3.15 IFD of pipeline network system with two customers ........................ 68
Figure 3.16 Pipeline network with two CSs and two customers ............................ 69
Figure 3.17 IFD for pipeline network with two CSs .............................................. 71
Figure 3.18 Flow chart of the TPNS simulation model based on Newton-Raphson
scheme .................................................................................................................... 75
Figure 3.19 Gunbarrel pipeline network system with two CSs .............................. 77
Figure 3.20 Branched pipeline network system ..................................................... 79
Figure 3.21 Looped pipeline network system ........................................................ 82
Figure 3.22 Part of the existing pipeline network system ...................................... 88
Figure 3.23 Proposed scheme for incorporating the age of the pipes in flow
equations ................................................................................................................. 97
Figure 3.24 The effect of length of service of the pipe on friction factor .............. 98
Figure 4.1 Convergence of pressure variables (case 1A) ..................................... 109
Figure 4.2 The convergence of flow variables (Case 1A) .................................... 110
Figure 4.3 The convergence of pressure variables (case 2A) ............................... 111
Figure 4.4 The convergence of flow variables (case 2A) ..................................... 112
Figure 4.5 Pipeline network with three CSs serving four customers ................... 113
Figure 4.6 IFD for pipeline network with three CSs and four customers ............ 114
Figure 4.7 Convergence of pressure variables (case 1B) ..................................... 117
Figure 4.8 Convergence of flow variable (case 1B) ............................................ 117
Figure 4.9 Comparison of nodal pressures ........................................................... 120
xx
Figure 4.10 Convergence of pressure (case 2B) ................................................... 122
Figure 4.11 Convergence of main flow variables (case 2B) ................................ 122
Figure 4.12 Convergence of lateral flow variables (case 2B) .............................. 123
Figure 4.13 Convergence of P1 for the first ten iterations for the initial estimations
(case 3B) ............................................................................................................... 126
Figure 4.14 Convergence of P2 for the first ten iterations for the initial estimation
(case 3B) ............................................................................................................... 126
Figure 4.15 Convergence of Q1 for the first ten iterations for the initial estimations
(case 3B) ............................................................................................................... 127
Figure 4.16 Pipeline networks with 10 pipes and two compressor stations [7] ... 128
Figure 4.17 Comparison of flow rates between TPNS simulation and results of
Osiadacz [7] .......................................................................................................... 131
Figure 4.18 Comparison of nodal pressures between TPNS simulation and results
of Osiadacz [7] ...................................................................................................... 132
Figure 4.19 Convergence of P1 for the first ten iterations for the initial estimations
.............................................................................................................................. 137
Figure 4.20 Convergence of P2 for the first ten iterations for the initial estimations
.............................................................................................................................. 138
Figure 4.21 Convergence of Q1 for the first ten iterations for the initial estimations
.............................................................................................................................. 139
Figure 4.22 Convergence of the main flow variables ........................................... 140
Figure 4.23 Discharge pressure variations with flow for various speeds ............. 141
Figure 4.24 Compression ratio variations with flow for various speeds .............. 142
Figure 4.25 Variation of power consumption with flow for various speed .......... 143
Figure 4.26 Convergence of pressure variables (Case 1C) ................................... 145
xxi
Figure 4.27 Convergence of flow variable (Case 1C) .......................................... 145
Figure 4.28 Convergence of pressure variables (Case 2C) .................................. 147
Figure 4.29 Convergence of main flow variables (Case 2C) ............................... 148
Figure 4.30 Convergence of lateral flow variables (Case 2C) ............................. 148
Figure 4.31 Effect of liquid holdup on nodal pressures ....................................... 149
Figure 4.32 Effect of liquid holdup on flow through main pipes ......................... 150
Figure 4.33 Convergence of pressure variables with corrosion (Case 3C) .......... 151
Figure 4.34 Convergence of flow variable with corrosion (Case 3C) .................. 152
Figure 4.35 Variation of flow with age of the pipe .............................................. 154
Figure 4.36 Variation of compression ratio with age of the pipe ......................... 155
Figure 4.37 Variation of the power consumption with age of the pipes .............. 156
Figure 4.38 Convergence of pressure variables (Case 4C) .................................. 157
Figure 4.39 Convergence of flow through main pipes (Case 4C) ........................ 157
Figure 4.40 Convergence of flow through lateral pipes (Case 4C) ...................... 158
Figure 4.41 Nodal pressures for different ages of pipes ....................................... 159
Figure 4.42 Flow through pipes for different ages of pipes ................................. 160
Figure 4.43 Convergence of pressures for non-isothermal conditions ................. 162
Figure 4.44 Convergence of flow variables under non-isothermal conditions .... 162
Figure 4.45 Convergence of some of the temperature variables .......................... 163
Figure 4.46 Simplified version of the modeling process [95] .............................. 165
Figure 4.47 Variations of flow rate with number of compressors ........................ 170
Figure 4.48 Variation of compression ratio with number of compressors ........... 171
Figure 4.49 Variation of power consumption with number of compressors ........ 172
xxii
Figure 4.50 Variation of main flow rate with source pressure ............................. 174
Figure 4.51 Variation of compression ratio with source pressure ........................ 175
Figure 4.52 Variation of power consumption with source pressure ..................... 176
xxiii
NOMENCLATURES
41 AA − Constants for normalized compressor head equations
PA Cross-sectional area of pipe
AGA American gas association
41 BB − Constants for normalized compressor efficiency equations
BHP Break horse power
BSCFD Billion standard cubic feet per day
C1, C2 Constants used to name pipes
PC Specific heat at constant pressure
CR Compression ratio
CS Compressor station
vC Specific heat at constant volume
D Diameter
iD Customer located at i
ijD Diameter of pipe joining node i and j
lD Volumetric flow rate of load pipe l
DP Dynamic programming
E Pipeline efficiency
f Darcy’s friction factor 'f Fanning friction factor
F Transmission factor
g Acceleration due to gravity
G Gas gravity
GA Genetic algorithm
GRG Generalized reduced gradient
h Elevation height
xxiv
H Adiabatic head
LH Liquid holdup
IFD Information flow diagram
k Specific heat ratio
ijK Pipe flow resistance between pipe node i and j
L Length
ijL Length of pipe joining node i and j
LNG Liquefied natural gas
M Mass flow rate
fM Total mass flux
GM Mass flow rate for the gas
LM Mass flow rate for the liquid
MINLP Mixed integer non linear programming
MMSCMD Million metric standard cubic meters per day
MMSCFD Million standard cubic feet per day
n Rotational speed of the compressor, exponent
cn Number of compressors
NDP Non sequential dynamic programming
jn Number of junctions within the pipeline network system
ln Number of loops within the pipeline network system
NLP Non linear programming
NP Number of unknown pressure variables
NQ Number of unknown flow variables
sn Number of compressor stations within the pipeline network system
NT Number of unknown temperature variables
NTotal Total number of unknown variables
P Pressure
avP Average pressure
xxv
sP , dP Suction, discharge pressure of the compressor
DP Demand pressure
nP Standard pressure condition
jq Volumetric flow rate of outgoing pipe j from the node
Q Volumetric flow rate
CiQ Gas flow rate to customer i
iQ Volumetric flow rate through pipe i
mQ Volumetric flow rate of the mixture
nQ Volumetric flow at standard conditions
r Roughness of the pipe as function of age
R Gas constant
airR Gas constant for air
Re Reynolds number
mRe Mixture Reynolds’s number
SA Simulated annealing
SS Successive substitution
PS Perimeter of the pipe
SCADA Supervisory control and data acquisition
t Number of incoming pipe to a node
T Temperature of gas
sT Average soil temperature
ST Temperature of the gas at the suction side of the compressor
ijT Temperature constant between upper node i and down node j
nT Standard temperature conditions
TPNS Transmission pipeline network system
u Number of outgoing pipes from a node
U Overall heat transfer coefficient
xxvi
Gv Velocity of the gas
Lv Velocity of the liquid
mv Velocity of the mixture
w Number of load pipes from a node
x Mass fraction (quality) X Vector representing unknown variables y Age of the pipe
Z Gas compressibility
1Z , 2Z Suction, discharge side compressibility of gas
Greek letters
α Void fraction
η Efficiency
Lλ No-slip liquid holdup
Gμ Viscosity of the gas
Lμ Viscosity of the liquid
mμ Viscosity of the mixtures
Gυ , Lυ specific gravity of the gas, liquid
θ Inclination angle
ρ Density
Gρ Density of the gas
Lρ Density of the liquid
NSρ The no-slip density of the mixture
mρ Density of the mixtures
wτ Shear stress at the wall of the pipe
xxvii
PREFACE
This thesis is submitted as a requirement for the degree of Doctor of Philosophy
in Mechanical Engineering for the author. It consists of the research work conducted
in the area of simulation of natural gas transmission from July 2006 to March 2010.
The problem was started from the observation of the congestions that usually
happened at natural gas distribution centers. The revision of the various literatures on
natural gas transportation showed that the distribution centers are the last system and
involved less volume of gas. It was observed that the problems occurred on the
transmission part of the pipeline network system. The main issues in transmission
pipeline network system (TPNS) were the determination of the nodal pressures,
flows and temperature variables which were used to evaluate the performance of the
system.
A simulation model was developed based on performance characteristic of
compressors and the principles of conservation of energy and mass of the system to
analyze TPNS with non-pipe elements for various configurations. The TPNS
simulation model analyzes single phase gas flow and two-phase gas-liquid flow. The
simulation model also takes into account temperature variation and age of the pipes.
The model was designed for the evaluation of the unknown pressure, flow and
temperature variables for the given pipeline network. The two solution schemes
developed were iterative successive substitution scheme for simple network
configurations and a generalized solution scheme which adopt Newton-Raphson
algorithm for complex network configurations. The results of TPNS simulation
model were compared with Newton loop-node method based on looped pipeline
network configurations and an exhaustive optimization technique based on gunbarrel
pipeline network configuration. For both cases, the results indicated that the model
is able to provide solutions similar to the compared models.
xxviii
Based on the results from the simulation tests and validation of the model, it is
noted that the developed TPNS simulation model could be used for performance
analysis to assist in the design and operations of transmission pipeline network
systems.
1
CHAPTER 1
INTRODUCTION
1.1 Background
Natural gas is becoming one of the most widely used sources of energy in the
world due to its environmental friendly characteristics. Usually, the location of
natural gas resources and the place where the gas is needed for various applications
are far apart. As a result, the gas has to be moved from deposit and production sites
to consumers either by trucks in the form of liquefied natural gas (LNG) or through
pipeline network systems. As reported in [1], short distances gas transportation by
pipelines is more economical than LNG transportation. The LNG transportation
incurs liquefaction costs irrespective of the distance over which it is moved. As a
result, the development of transmission pipeline network system (TPNS) for natural
gas is a key issue in order to satisfy the ever growing demand from the various
customers.
When the gas moves by using the TPNS, the gas flows through pipes and various
devices such as regulators, valves, and compressors. The pressure of the gas is
reduced mainly due to friction with the wall of the pipe and heat transfer between the
gas and the surroundings. Compressor stations are usually installed to boost the
pressure of the gas and keep the gas moving to the required destinations. It is
estimated that 3 to 5% of the gas transported is consumed by the compressors in
order to compensate for the lost pressure of the gas [2]-[4]. This is actually a huge
amount of gas especially for the network transmitting large volume of gas. At the
current price, this represents a significant amount of cost for the nation operating
2
large pipeline network system. For instance, considering the U.S. TPNS, Wu [2]
indicated that a 1% improvement on the performance of the transmission pipeline
network system could result a saving of 48.6 million dollars. Carter [3] also
presented that the cost of natural gas burned to power the transportation of the
remaining gas for the year 1998 is equivalent to roughly 2 billion dollars for U.S.
transmission system.
Investigation on various TPNS indicated that the overall operating cost of the
system is highly dependent upon the operating cost of the compressor stations which
represents between 25% and 50% of the total company’s operating budget as
discussed in [5] and [6]. Hence, compressor station is considered as one of the basic
elements in TPNS.
The main issues associated with both design and operating TPNS are minimizing
the energy consumption and maximizing the flow rate through pipes. Over the years,
numerical simulations of TPNS have been carried out in order to determine the
optimal operational parameters for given networks with various degrees of success
[2], [3], [7]-[9]. From the optimization perspective, the problem of developing an
optimal TPNS is nonlinear programming problem where the objective function is
typically nonlinear and non-convex, and some of the constraints are also nonlinear.
Different techniques are proposed in order to get the optimal parameters of TPNS by
either modifying the objective function or relaxing some of the constraints [10]-[15].
However, due to the complexity of the objective function and the constraints, the
determination of optimal parameters for TPNS is yet challenging from the
optimization perspective.
On the other hand, simulation has contributed significant achievements in
analyzing the TPNS problems [7], [9], [14], [16]- [20]. TPNS simulation is used to
determine the design and operating variables of the pipeline network for various
configurations. The applications of simulation for TPNS can be summarized as
follows:
3
1. Generate and evaluate various configurations of TPNS in order to guide in
the selection of optimal system.
2. Assist in making decisions regarding the design and operations of pipeline
network systems. During the design process, simulation could assist in
selecting the structure of the network and the geometric parameters of the
pipes which satisfy the requirements. Furthermore, it also facilitates the
selection of sites where compressors, valves, regulators and other elements
should be installed.
3. Predict the behavior of TPNS under different operating conditions.
4. Plays a vital role in analyzing the existing TPNS to study how the system
responds for future variations in demand and supply. The effect of
additional customer sites, the addition of new pipes or compressor stations
on the existing system can also be studied with the aid of simulation.
TPNS in most countries consist of a large set of highly integrated pipe networks
operating over a wide range of pressures. An increase in demand for natural gas
enhanced the development of complex TPNS. The basic problems associated with
reasonable operations of TPNS are proper supply of gas to the consumers and low
system operating costs. Proper (optimum with regard to a certain criterion)
development of transmission pipeline network system, as well as its economically
rational exploitation, are only possible if simulation procedures are applied [7].
There are three types of gas transmission networks which include gunbarrel,
branched and looped pipeline configurations [6]. The complexity of the simulation
analysis depends on the extent of the pipeline network configurations (gunbarrel,
branched, looped, etc.), the nature of the gas (single phase dry gas, two-phase gas-
liquid mixture) and other factors such as temperature of the gas, the number of
sources of the gas (single source, multi-source) and internal pipe corrosion.
4
TPNS consists of pipes and non-pipe elements such as compressors, regulators,
valves, scrubbers, etc. The simulation of TPNS system without the non-pipe
elements is relatively easier to handle and developed by Osiadacz [7]. The addition
of non-pipe elements makes the simulation of TPNS more complex due to the
modeling of the non-pipe elements. More equations have to be added into the
governing simulation equations when the non-pipe elements are considered during
analysis. Compressor station is one of the main non-pipe components of gas
transmission system and considered as a key element.
One of the basic differences among TPNS simulation analysis models with non-
pipe elements is in the way compressor station is modeled during simulation. There
have been attempts reported by various researchers on modeling compressor stations
within the TPNS during simulation [7], [16], [18], [21]. One of the options is to
consider the compressor station as a black box by setting either the suction or
discharge pressures [21]. Only little information can be obtained to be incorporated
into the simulation model to represent the compressor station. The effect of
compressor station during simulation of TPNS has been incorporated by pre-setting
the discharge pressures [7], [16], [18]. However, the speed of the compressor, suction
pressure, suction temperature, and flow through the compressor were neglected
during the analysis. Even though there have been attempts reported regarding the
simulation of TPNS with non-pipe elements, there are issues that are not addressed.
1.2 Problem Statement
Generally, from the previous approaches on TPNS simulation, it is observed that
compressor station is considered as a black box by setting few parameters or its
effect is oversimplified during simulation. This is due to the fact that the addition of
non-pipe elements makes the TPNS simulation more difficult to analyze. As a result,
5
few attempts [2], [7], [9], [16], [18] and [21]. have been done to have a complete
TPNS simulation with all its components. Since compressor station is one of the
basic elements in TPNS, the detail incorporation of all its parameters, namely: speed,
suction pressure, discharge pressure, flow rates, number of compressors and suction
temperatures is essential for a complete simulation of gas pipeline networks.
As the age of the pipe increases, the roughness of the pipe increases due to the
accumulation of various elements around the internal surface of the pipe [22]. This
might results in decrease in performance of the TPNS system, i.e. lower flow rate
capacity and higher pressure drop [23]. Limited information are available in the
literatures regarding the roughness with the age of the pipe and the performance of
the system. It is beneficial to have a TPNS simulation model to evaluate the
performance of the system with the age of the pipes.
In the petroleum industry the transportation of gas and low loads of liquids
(usually less than 0.005 holdups) occurs frequently in transmission pipelines for both
onshore and offshore operations. The liquids are usually heavy hydrocarbon fractions
and water which may be introduced from several sources. Liquids from the
compression facilities, and treatment plants as well as products of condensation may
also accompany the gas during transportation [24]. The use of single phase flow
equation for the analysis of TPNS with two-phase mixtures might lead to
underestimation of the pressure drop and flow capacity of the system. As a result, a
TPNS simulation model with appropriate flow equations which take into
consideration for the existence of liquid in the system is very useful.
6
1.3 Research Objective
The main objective of this research is to develop a TPNS simulation model for
the analysis of the performance of pipeline network system incorporating compressor
characteristics, effect of two-phase flow and the age of the pipes.
In order to achieve the objective, the thesis addresses the following issues within
the developed methodology:
1. Identifying factors that should have been considered for the design
and operation of natural gas transmission pipeline network with non-
pipe elements.
2. Formulating detailed mathematical model for the simulation which
takes into account the pipeline configurations (gunbarrel, branched
and looped), compressor characteristics, nature of the gas (single
phase, two-phase gas-liquid), temperature of the gas and the age of the
pipes.
3. Developing iterative successive substitution solution scheme for
determining the unknown pressure and flow variables and using excel
based spreadsheet to assist in evaluating different scenarios of
operation of TPNS.
4. Developing Newton-Raphson solution scheme for determining the
unknown variables and using visual C++ code to help in evaluating
the different scenarios for the design and operation of pipeline
network system.
5. Validating the Newton-Raphson simulation model with the aid of
appropriate validation techniques which includes simulations,
7
comparison with previous models, and case study based on the
existing pipeline network system.
1.4 Research Scopes
The scopes of the research are summarized as follows.
1. Pipeline network system for moving gas is divided into three classes:
gathering, transmission and distribution system. Gathering pipeline
network system is responsible for collecting individual gases from gas
wells and storage system and move to the gas processing plant.
Transmission pipeline network system is used to move the gas from
gas processing plant and deliver to the distribution centers and large
industrial customers. The distribution network system is responsible
for routing of gases to individual customers. This research focuses on
the transmission system since all the gases from the gas processing
plant passes through this network system. Single source flow with one
pipe directing the gas to the compressor station is assumed as the gas
processing plant is the only source of gases to the transmission
system.
2. Development of TPNS simulation model to make performance
analysis for the gas pipeline networks involving flow capacity,
compressor ratio and power consumption for the system.
3. In a TPNS problem, the system can be modeled as steady state or
transient model depending on how the gas flow changes with respect
to time. In a steady state TPNS, the values characterizing the flow of
gas in the system are independent of time. On the other hand, transient
8
analysis requires the use of partial differential equations to describe
the relationships between parameters which make the problem more
difficult to analyze. In case of transient simulation, variables of the
system, such as pressures and flows, are function of time. This
research focuses on steady state system which is a common practice in
gas pipe line network system.
4. TPNS mainly consists of pipes and many other non-pipe devices such
as compressor stations (CS), valves and regulators. Although there are
many non-pipe elements in TPNS, CS is the key characteristics in the
network. This research focuses on addressing TPNS simulation with
CS as non-pipe element. Centrifugal compressor is assumed
throughout this study due to its frequent application in gas industry.
All the compressors within the compressor stations are arranged in
parallel.
1.5 Structure of Thesis
The remainder of the thesis is organized as follows.
Chapter 2 presents the literature review on the existing single phase optimization
and the simulation approaches for TPNS problem. It also includes the review of the
various two-phase flow models, modeling temperature variations and internal
corrosion. The basic solution methods applied in TPNS simulation are also
discussed.
The third Chapter describes the methodology used to achieve the objective
described in Chapter 1. It includes the detail description of the basic mathematical
formulation for the governing simulation equations based on pipeline configurations,
9
nature of the gas, temperature of the gas and internal corrosion of the pipes. The
detailed description of the iterative solution schemes to get the unknown variables
are also presented in this Chapter. Furthermore, the applications of the TPNS
simulation model based on successive substitution and Newton-Raphson for
modeling various network configurations are also presented.
The results of the research are discussed in detail in Chapter 4. The results of
TPNS simulation model based on successive substitution and Newton-Raphson
scheme using various TPNS configurations are presented. Moreover, the simulation
model is also tested based on the existing pipeline network system. Results of the
simulation model compared with the previous models are also discussed.
Chapter 5 presents summary of the research and the main contributions derived
from this research. Issues requiring further study are also addressed in this Chapter.
10
CHAPTER 2
LITERATURE REVIEW
2.1 Introduction
For the last several decades, numerical simulations of TPNS have been carried
out at various stages of TPNS development like feasibly study, economic evaluation,
sizing, operations and monitoring of the network with various degree of success. The
simulation and optimization of TPNS have been done on the basis of different
assumptions. The process of TPNS simulation and optimization become complex
depending on the situations considered. The complexity arises from whether the
simulation or optimization model consists of single phase or multiphase flow
conditions, transient or steady state conditions, isothermal or non-isothermal
conditions and single pipe or pipeline networks conditions. Auer [25] discussed the
importance of building useful simulation model. The determining factor on how
useful the simulation model is its ability to predict the parameters of TPNS under a
wide range of conditions.
This chapter discusses the review of the most relevant literatures on the area of
single phase flow optimization by introducing the brief history of pipeline
development. The review of single phase flow simulation of pipeline networks with
non-pipe elements and with only pipe elements are also discussed. Simulation
models which consider two-phase flow analysis, temperature variations and internal
corrosion are also reviewed. The last part of this Chapter discusses the review of the
solution schemes applied for pipeline network simulations.
11
2.2 Optimization of Gas Pipeline Networks
With increased applications of natural gas in power, industrial, commercial and
residential customers, the demand for natural gas is increasing for the past decades
[26]. The pipeline network gets more complex as the demand for natural gas
increases. This enhanced the importance of optimum design of pipeline network for
natural gas distribution system.
The first pipelines were built in the late 1800s to transport coal gas through cast
iron and lead pipes for street lighting. Long distance, high pressure pipeline began
operating in the United States in 1891 [27]. The development of seamless steel pipes
allowed transmission of gas at high pressure and greater quantities [28].
The world pipeline network expanded rapidly when it became apparent that
pipelines were an efficient, economic way to move oil, gas, and products to
consumers. The main contributing factors for the expansion of pipelines network
system were large discoveries of oil and gas in remote areas where there is only little
local demand. Pipelines were needed to move these supplies to market [29].
One of the main concerns with both design and operating TPNS is minimizing
the energy consumption by the system while satisfying the specified delivery
requirements throughout the system. The mathematical formulation for energy
minimization of gas transmission network is given by Mercado [6]:
Minimize ( )1.2),,(),(∑∈ snji
jiijij PPQg
Subject to
( )2.20111
=−− ∑∑∑===
w
ll
u
jj
t
ii DqQ
( )3.2222ijijji QKPP =−
[ ] ( )4.2, Ui
Lii PPP ∈
12
( ) ( )5.2,, ijjiij DPPQ ∈
where ijD represent the feasible operating domain of the compressor stations
and ),,( jiijij PPQg is the fuel consumption function at the compressor station.
In equation (2.1), the term ijQ represents the flow through the compressor station
while iP and jP are the suction and discharge pressures. Equation (2.2) is the mass
balance equation at each junction where iQ is the flow through all incoming pipes,
jq is the flow through all outgoing pipes, and lD is the load flow at each junction
node. Equation (2.3) is the pressure drop equation relating the pressures at the
upstream and downstream. The term ijK is the pipe resistance which is a function of
pipe physical properties. Equation (2.4) represents the upper and lower pressure
limits at each node while equation (2.5) is the constraints introduced for compressors
to work within the feasible domain.
From the optimization perspective, the problem of developing an optimal TPNS
is a nonlinear programming problem where the objective function, equation (2.1), is
typically nonlinear and non-convex, and some of the constraints, equation (2.3) and
equation (2.5), are also nonlinear. The problem is very difficult due to the non-
convex nature of both the objective function and the feasible region. However,
various researchers attempted different techniques in order to get the optimal
parameters of TPNS by either modifying the objective function or relaxing some of
the constraints. The following sections discuss the most commonly used techniques
in TPNS optimization.
2.2.1 Dynamic Programming
One of the techniques for TPNS optimization which has been widely applied in
previous studies since the late 1960s is dynamic programming (DP). DP approach
[30] and [31] for pipeline network optimization was first applied by Wong and
13
Larson [8]. They applied the method for fuel cost minimization of straight line
system and used recursive formulation. The gunbarrel system (Figure 2.1), which is
basically straight line type pipeline network system, possesses an appropriate serial
structure to be solved by DP.
Figure 2.1 Gunbarel structure transmission pipeline network adapted from [8]
Wong and Larson [32] then extended the principle of the application of DP from
gunbarrel type gas transmission system to a more general single source tree
structured pipeline network system. They decomposed the tree structured network
into sequential one dimensional DP problem in order to optimize the network.
For the looped structure, DP has been applied to simplified cases. Carter [3]
proposed a DP approach on more general structure with flow rates being fixed. He
developed a non sequential dynamic programming (NDP) technique which allows
pure DP to general branched and looped systems. In NDP, rather than attempting to
formulate DP as a recursive algorithm, two or three connected compressor or
regulator elements are replaced by a virtual composite element that behaves just like
its components operating in an optimal manner. Three types of composition
operations were developed in order to reduce complex system to simple one.
DP has been also combined with other techniques in order to optimize TPNS.
Ríos-Mercado et al. [6] and [33] proposed a heuristic solution procedure for fuel cost
minimization on TPNS with cyclic (looped) network configurations. The heuristic
procedure was based on a two stage iterative. At first stage, gas flow variables are
fixed and optimal pressure variables are found via DP. At a second stage, pressure
14
variables are fixed and an attempt is made to find a set of flow variables that improve
the objective by exploiting the basic network structure.
Borraz-S'anchez and R'ıos-Mercado [4] developed a hybrid meta-heuristic
solution procedure for fuel cost minimization of TPNS with cyclic topology on the
basis of DP. The heuristic methodology was based on two stage iterative procedure.
In the first stage, gas flow variables are fixed in each network arc and optimal
pressure variables in each network node were found via NDP approach developed by
Carter [3]. In the second stage, pressure variables are fixed and a short-term memory
tabu search procedure was used for guiding the search in the flow variable space.
Empirical evaluations have been made on TPNS with different configurations and
the results were compared with NDP and Generalized Reduced Gradient (GRG)
solution technique. The algorithm is capable of obtaining solution for the instances
considered. However, it is computationally expensive when compared to NDP and
GRG.
Kim [5] proposed a heuristic solution procedure for minimizing fuel cost
consumption of TPNS based on DP. The algorithm used DP as part of an iterative
procedure that modifies flow and pressure separately in an attempt to find a better
solution. The solution procedure is an iterative process. First, flow variables were
fixed by flow modification, and then DP was used to find an associated set of
pressure variables. Due to the non convexity of the objective function and the
constraints set, there is no theoretical guarantee of convergence to a local or global
minimum.
The major reason DP is attractive for optimization of TPNS problem is because it
is a general model to apply for this types of problems. Nonlinearities in the system
constraints and performance criterion can easily be handled, and constraints on both
decision and state variables introduce no difficulties. However, the application of DP
is limited to linear (gun barrel) or tree topologies. Furthermore, the computation
15
process increases exponentially with the dimension of the problem as presented in
Carter [3].
2.2.2 Mathematical Programming
Although non-linearity of the TPNS problem makes it very difficult to apply the
mathematical programming technique as a solution procedure, several researchers
have attempted to develop optimal TPNS using mathematical programming
approach. The limitation in mathematical programming is that, the nonlinear
relationships that exist either in the objective function or constraints have to be
approximated by linear relationships. This could result in a deviation from the
original problem.
Mőller [11] formulated the optimization of TPNS problem as mixed integer
linear programming. The non-linearity in both the objective function and the
constraints functions are approximated by piecewise linear functions. A separation
and branch-and-bound algorithm were developed in order to get rid off binary
variables and to fasten the calculations. The algorithm was tested on large networks
to obtain the optimal values for the TPNS. However, the non-linearity and complex
relationship between flows and pressures of pipes and compressors make the
modeling difficult for the optimization of TPNS.
Wolf and Smeers [12] proposed piecewise linear programming method to solve
the gas transmission problem. The authors formulated the problem in gas
transmission as cost minimization problem subject to the nonlinear flow pressure
relationships, material balance and pressures bounds. The solution method was based
on piecewise linear approximation of nonlinear relationships that exist during the
problem formulation. The approximated problem is solved by an extension of the
simplex method.
16
Ruz et al. [13] developed modular, constrained based model for TPNS in order to
answer questions concerning the supply, demand and transportation in the context of
optimization. The goal of the model was precise estimation of its transport capacity
to be used in a wider logistic model. The planning and scheduling for gas supply
used a mixed integer optimization model and consists of a set of interconnected
modules. Each module implements the behavior of a physical device of the gas
transmission network. The model approximates the nonlinear relationships for pipes
and compressor stations by linear functions.
Hoeven [14] formulated the gas transmission problem as constrained network
simulation model by linearizing the nonlinear relationships. The method was based
on modeling each of the elements of TPNS (compressors, pipes, valves, reducer, etc)
as constraints and relationships. The linearized equations were finally solved using
an iterative scheme.
Pratt and Wilson [10] proposed a mathematical programming approach for
optimization of the operation of TPNS. Their approach solves the nonlinear
optimization problem iteratively by linearizing the gas flow equations, constraints
and objective functions to give a linear constrained problem. The linear constrained
problem then optimized by mixed integer linear programming and solved using
branch and bound algorithm technique. The re-linearization about the optimum will
repeat until a convergence is obtained for the system. They included compressor fuel
cost and cost of flows from sources as the objective functions.
Percell and Ryan [34] proposed an algorithm using generalized reduced gradient
scheme for minimizing the fuel consumption problem for TPNS. Being a method
based on gradient search, there is no guarantee for a global optimal solution,
especially when there are discrete decision variables.
Abbaspour [9] developed a solution procedure for tackling pipeline operation
problem using simulation based optimization. The first step is to device an analysis
17
scheme that provides the simulation support required by the optimization. By gaining
some information from the simulation, the problem of gas transmission operation
was modeled as non linear programming (NLP) and solved with the sequential
unconstraint minimization technique. Since the method is based on gradient search,
there is no guarantee for global optimal solution. Furthermore, the optimization is
done at the station level rather than at network level.
Baumann et al. [15] presented a gas network optimization program (GassOpt) as
a decision support to the design of pipeline network system. The program has been
developed for transport analysis and planning purpose in order to calculate optimal
routing of natural gas in pipeline network system. The simulation program is
developed based on linear programming and optimizing engine.
Osiadacz and Gorecki [35] and Wu et al. [36] proposed solution algorithms that
tries to minimize the cost of development of pipes. The algorithms are based on
approximation of the non-linear relationships with linear constraints. A statistical
modeling technique for the design and operation of TPNS based on Monte Carlo
simulation has been proposed by Chmilar [37]. The approach is based on performing
successive approximations or estimations of the system load (flow) and capability to
generate an overview of how the system can be expected to perform.
2.2.3 Expert System
Different scholars [38]–[41] have used expert systems approach in order to
optimize the operations of TPNS. One of the main challenges that exist in applying
the expert system for optimizing TPNS operation is knowledge extraction. The
knowledge that may be extracted through interviews of experts from the natural gas
industry might not be sufficient enough to actually represent the real situations.
Sun et al. [38] proposed expert system to enhance the decision making abilities
of the dispatcher in order to optimize the natural gas operation. Sun et al. [39]
18
developed an integrated approach, both expert systems and operations research
techniques to model the operations of the gas pipelines. In the integrated model, the
expert system performs the decision to inform the dispatcher on the current system
linepack level with the associated control action to be issued and how much
horsepower should be added/reduced to satisfy future demand. On the other hand, the
decisions to turn on/off which compressor are handled by the mathematical model
adopted in the operations research. The model was tested on branched TPNS with
two compressor stations serving two customers. The optimization program
exhaustively tests all possible combinations of the compressors which increase the
complexity of the decision when the number of compressors increases.
Uraikul et al. [40] and [41] proposed an expert system as a decision support tool
to optimize natural gas pipeline operations. The expert system was developed as a
feasibility study on applicability of expert system technology to pipeline
optimizations. The three major tasks that the proposed expert system performs in
order to provide decision support to the operators are:
• Determining whether the linepack is high, low or enough,
• Suggesting the amount of break horsepower needed to increase or decrease
the linepack, and
• Recommending a compressor unit to turn on or off.
The proposed technique highly depends on the knowledge acquisition which has
been derived from historical data analysis, heuristic knowledge from expert operators
from the natural gas industry and a computer simulation model. Furthermore, the
expert system in its current version does not adequately deal with uncertain
information.
19
Chebouba et al. [42] proposed ant colony optimization algorithm to solve how to
operate TPNS efficiently. The algorithm was tested on straight line pipeline (gun
barrel) network system with five compressor stations. An optimal operating policy
for TPNS was developed for two different flow rates. The results were compared
with that of DP approach and it was reported that the results obtained were similar in
most of the cases with less computational time.
Mora and Ulieru [43] used genetic algorithm to reduce the energy used to
operate compressor stations in TPNS problem. The solution methodology was based
on two stages. The first stage of the methodology was the use of genetic algorithm
for speeding up the searching process to provide a solution in a timely manner. Each
candidate solutions generated by the search algorithm has been evaluated by
hydraulic model that simulates the steady state gas flow in the TPNS to obtain the
reaction of the system at specific control nodes and determine the feasibility of the
given solution at the second stage. Genetic algorithm has been attempted for gas
transmission network [44] and [45].
2.2.4 Analytical Hierarchical and Network Reduction
Hierarchical structure and network reduction techniques have been suggested by
various researchers as solution procedures when it is difficult to solve the gas
transmission problem in an integrated way. Analytical hierarchical approach involves
the decomposition of the gas transmission network problem into pipeline network
level and compressor station level. Some degree of success has been achieved as far
as the optimization of the compressor station subproblem is concerned. However, as
stated by Mercado [6], these approaches have limitations in globally optimizing the
minimum cost.
Osiadacz [46] developed an algorithm for optimal control of gas network based
upon hierarchical control technique. The network was divided into physically small
20
subsystems by imposing a constraint of incorporating at least one operating
compressor in each subsystem. Local problems were solved using gradient
technique. The subsystems were coordinated using “goal coordination” method to
find the overall optimum.
Wu et al. [47] developed an algorithm for optimal operation of pipeline network
system based on relaxation of the objective function (fuel cost minimization) and the
constraint (the feasible domain of the compressor station). The authors generated
four different problems for the given network and showed the differences that exist
by solving the original problem and,
• The modified problem by relaxing the feasible domain of compress stations,
• The modified problem by relaxing the fuel cost function, and
• The modified problem by relaxing both the feasible domain of the
compressor stations and the fuel cost function at the same time.
The authors considered three network problem instances. The first two problems
were a kind of network where one can get the optimal solution through exhaustive
search method. It has been reported that, lower bound solution gave an approximated
solution with good relative optimality gap for the two problem instances considered
in the paper. However, no comparison was made about the approximation of the
lower bound for the third problem as it is difficult to get the optimal solution.
Rios-Mercado et al. [48] proposed a network reduction technique for TPNS
optimization problem based on a combination of graph theory and non-linear
functional analysis. It has been reported that, the reduction technique reduced the
problem size significantly without disrupting its mathematical structure. However, no
comparison has been made to justify the applicability of the procedure and its
effectiveness compared to the existing approaches. Mohring et al. [49] presented a
21
methodology of automated model reduction for TPNS. The method is based on
computer algebra to compose automatically the model equation for different
components of TPNS (pipe, compressor, regulator, etc…). It has been reported that
the model complexity could be reduced significantly when the method is applied for
the network. However, checking of whether the parameter is within the predefined
errors is also a time consuming process.
2.3 Simulation of Gas Pipeline Networks
There are various definitions for the term simulation. One of the definitions
which is adopted in this thesis is the definition by Chung [50]. Simulation modeling
and analysis is the process of creating and experimenting with a computerized
mathematical model of a physical system. A system is defined as a collection of
interacting components that receives input and provides output for some purpose.
The simulation modeling and analysis of different types of systems are conducted
for the purposes of gaining insight into the operation of the systems. The simulation
analysis can be used for developing operating or resource policies to improve system
performance and testing new concepts. It is also used for gaining information without
disturbing the actual system and generating sample operations of the system with
different patterns and configurations [50].
Simulation of gas pipeline networks could be used to determine pressure, flow
and temperature variables of the network under different conditions. As a result,
based on the variables obtained, simulation could assist in the decisions regarding the
design and operation of the real system. Osiadacz [7] stated that at the stage of
designing TPNS, simulation could help to select the structure of the network and the
geometric parameters of the pipes which satisfy supply and demand requirements. At
22
the stage of operating TPNS, simulation could help to make various scenario
analyses in order to guide for optimal operations.
The complexity of simulation of gas pipeline network depends on the extent of
the system. Some authors performed pipeline network simulation analysis by
neglecting the non-pipe elements of the network or oversimplify their effect. The
following sections discuss the most related literatures on pipeline network simulation
with or without the non-pipe elements.
2.3.1 Simulation of Gas Networks without Non-pipe Elements
Pipeline network simulation without non-pipe elements like compressor stations,
valves, regulators is less challenging as it involves only pipes. The number of
equations and the type of equations are less compared to the simulation of pipeline
network elements with all its components. For instance, assuming constant
temperature of the gas, only pipe flow equations and mass balance equations are
involved during the simulation of the pipeline network system without non-pipe
elements. The simulation of pipeline network system without non-pipe elements are
developed by Osiadacz [7] based on graph theory.
One of the governing equations during the simulation of TPNS without non-pipe
elements is pipe flow equation. The flow of gas through pipes can be affected by
various factors such as the gas properties (specific gravity, viscosity, compressibility
and density), friction factor and the geometry of the pipes.
The relationships between the upstream pressure, downstream pressure and flow
of the gas in pipes for steady state conditions can be described by various equations
[7] and [23]. Friction factor is one of the main reasons for the variation of flow
equations.
23
In this thesis, general flow equation is adopted due to its frequent application in
gas industry [2], [5], [7] and [8] to describe the relationships between pressure
difference and gas flow in pipes. The general flow equation for the steady flow of
gas in a pipe is derived from Bernoulli’s equation. For an inclined pipe shown in
Figure 2.2, with length L and diameter D , the pressure drop between upstream node
1 and downstream node 2 of the pipe can be expressed using general flow equations
as [7]:
( )6.2264 222
522
22
1 ghTZRGPL
TPQ
DRfGZTPP
air
av
n
nn
air+⎟⎟
⎠
⎞⎜⎜⎝
⎛=−π
Hence, from equation (2.6), the flow nQ is given by
( )( )7.2
2
64
52
22
212
fGZTL
DTZR
GghPPP
PTRQ
air
av
n
nairn
⎪⎭
⎪⎬⎫
⎪⎩
⎪⎨⎧
⎥⎥⎦
⎤
⎢⎢⎣
⎡−−
×⎟⎟⎠
⎞⎜⎜⎝
⎛=
π
If the pipe is horizontal, the elevation term )/(2 2 TZRGghP airav is zero and equation
(2.7) reduces to
( )8.2])[( 522
21
fGZTLDPP
PTCQ
n
nn
−×=
where ⎟⎟⎠
⎞⎜⎜⎝
⎛=
64
2airRC π
24
Figure 2.2 Flow of the gas though pipe adapted from [7]
Several flow equations are in use in the gas industry, all of which are
modification of the general flow equations. The differences between them depend
mainly on what expression is assumed for the friction factor f . Table 2.1 shows the
summary of the various flow equations relating the upstream node i and downstream
node j pressures for horizontal pipes. Note that for determination of the pipe flow
resistance ijK , 589.0=G , KT 0288= , kPaPn 101= , KTn0288= , KkgJRair /5.287= ,
95.0=Z are assumed throughout all equations.
25
Table 2.1 Gas flow equations and range of applications adapted from [7] and [23]
Name of Equation Range of pressure Equation
General > 700 kPa 222
ijijQKPP ji =−
51069.3 4
D
fLijK ×=
Panhandle ‘A‘ >700 kPa 854.122ijijji QKPP =−
854.4251067.1
DELKij ×=
Weymouth >700 kPa 222ijijji QKPP =−
113.7281015.8
DELKij ×=
Polyflo 75 to 700 kPa 848.122ijijQKPP ji =−
848.4251079.2
DELKij ×=
Lacey 0 to 75 kPa 2
ijijji QKPP =−
51080.1 6
D
fLKij ×=
In addition to the pipe flow equations, mass balances provide the remaining basic
equations of the governing equations for the simulation of TPNS without non-pipe
elements. The mass balance equations can be obtained based on the principle of
conservation of mass at each junction of TPNS as presented in equation (2.2).
In practical situations, the non-pipe elements like compressor stations are usually
encountered in gas transmission network systems. As a result, it is important to
address the effect of these elements into the simulation model. This study focuses on
the simulation of TPNS with non-pipe elements.
26
2.3.2 Simulation of Gas Networks with Non-pipe Elements
The addition of non-pipe elements makes the simulation of pipeline network
system more complex and difficult to handle. More equations have to be added into
the governing simulation equations when the non-pipe elements are considered
during simulation analysis. Compressor station is one of the main non-pipe
components of any gas transmission system and considered as a key elements [5] and
[6]. It adds energy to the gas in order to overcome the frictional losses and to
maintain the required delivery pressures and flows. Bloch [51] and Hanlon [52]
discussed the applications and basic principles of compressors.
One of the basic differences among TPNS simulation models is the way how
compressor station is modeled during simulation. Several attempts have been made
by various researchers on modeling compressor station within the pipeline network
system during simulation. One of the option which is suggested by Letniowski [21]
is to consider the compressor station as a black box by setting either the suction or
discharge pressures. In this case, only little information can be obtained to be
incorporated into the simulation model. Compressor station provides mass balance
equation and information regarding the suction pressure which are the two equations
added to the remaining pipe flow equations to have a complete model for the
network.
Osiadacz [7] incorporated the effect of compressor station for simulation of
pipeline network system by presetting the discharge pressures at each compressor
stations. The procedure started by making a cut-off at compressor stations. The nodes
representing compressor stations in the node data table become input nodes and the
compressor station output is denoted by an additional node. After the cut-off is made,
all nodes with preset discharge pressures become reference nodes. Reference nodes
are nodes where the pressure is known in advance of the simulation.
27
Nimmanonda et al. [16] developed a computer aided model for design of a
simulation system for TPNS which enhance iteration between the user and the
system. The simulation design program obtains the input data (pipeline pattern,
natural gas constituents, number of compressors at each station, compressors’
horsepower, pipe size, range of operating pressure, and customers’ consumption) to
generate the variables of gas properties, pressures, and flow rate of the entire pipeline
system. Continuity, mass balance, and energy balance equations were considered to
simulate the pipeline system. The simulation system can generate sample operations
of the TPNS with different configurations. However, the authors did not mention
how the compressor stations are handled during the simulation. Only TPNS without
loop was considered during the simulation model. Furthermore, the simulation model
is developed for specific geographic location which limits its applicability to the
various weather conditions.
Nimmanonda [18] developed computer aided simulation model for natural gas
pipeline network system operations where the compressor stations are modeled based
on the relationships between flow rate, break horse power (BHP), and compression
ratio. The relationship between compression ratio, BHP, and flow was developed for
three flow categories. The model needs presetting either the suction or discharge
pressure and also the speed of the compressors was not taken into considerations.
The modeling of compressor station integrated in TPNS is very difficult as the
compressor stations may contain various compressors with different arrangements.
However, there have been various attempts conducted to optimize and simulate the
operation of only the compressor stations [19], [53]-[58]. The most commonly used
compressors in gas industries are centrifugal and reciprocating. Centrifugal
compressor is assumed throughout this study due to its frequent application in gas
industry.
28
Usually, the data related to compressor are available in the form of compressor
performance characteristics as shown in Figure 2.3. In order to integrate the
characteristics of the compressor into the simulation model, it is necessary to
approximate the characteristics map with mathematical models. The basic
quantities related to a centrifugal compressor unit are inlet volume flow rate Q ,
speed n , adiabatic head H , and adiabatic efficiency η . The mathematical
approximation of the performance map of the compressor can be done based on the
normalized characteristics.
29
Inlet volume flow (m3/hr)
Isentropic Head (KJ/kg)
Figure 2.3 Typical performance map of centrifugal compressor [59]
The three normalized parameters which are necessary to describe the
performance map of the compressor includes, 2H/n , Q/n and η [60]. Based on the
normalized parameters, the characteristics of the compressor can be approximated
either by two degree [9] or three degree polynomials [2]. Three degree polynomial
which gives more accurate approximation is used in this study. Applying the
principles of polynomial curve-fitting procedures for each compressor, the
30
relationship among the basic normalized parameters can be best described by the
following two equations:
( ) ( ) ( ) ( )9.234
2321
2 nQAnQAnQAAnH +++=
( ) ( ) ( ) ( )10.234
2321 nQBnQBnQBB +++=η
where 4321 A,A,A,A and 4321 B,B,B,B are constants which depends on the unit
to define a particular compressor.
Equation (2.9) was used as basis for modeling the governing simulation equation
to represent the compressor stations. The detail modeling of compressor stations
within the TPNS will be discussed in Chapter 3.
2.4 Simulation Models with Two-phase, Temperature and Corrosion
As discussed in the previous section, one of the variations among the simulation
models is whether the model addresses the non-pipe elements or not. Apart from the
elements of the pipeline network system, TPNS simulation models have variations
among themselves based on nature of the gas. Some models assume only single
phase dry gas during the modeling of the flow equations. On the other hand, some
models perform the simulation based on two-phase gas-liquid mixture assumption.
The other variation among simulation models for gas transmission systems comes
from assumption of temperature of the gas and internal pipe corrosion during
modeling. This section presents the review of related literatures on the effect of two-
phase gas-liquid mixtures analysis, temperature, and internal corrosion during the
simulation of TPNS.
31
2.4.1 Two-phase Flow Models
Usually, in a petroleum industry both gas and low load liquids might occur in
natural gas gathering and transmission pipelines network system [61] and [62]. The
sources of the accompanying liquids could be compression facilities, treatment
plants, and/or condensation of the gas during transportation process. Several
researches have been conducted on modeling the effect of the liquids on the
transmission efficiency of the TPNS. The followings are the review of most related
literatures on the area of the transmission of gas and low load liquids.
The accompanying liquids during the transmission of gas will affect the
transportation efficiency of the system. Asante [24] suggested that, most gathering
pipelines (which typically have liquids loads up 560m3/Million m3 of gas) transport
fluids as multiphase components. On the other hand, for transmission pipelines,
where the liquid entrainment is usually less than 56m3/Million m3 of gas, most
pipeline companies typically employ “dry gas” models to predict the transport
capabilities of the system. In reality, the accompanying liquid may travel as a film or
may be distributed as dispersed droplets in the predominant gas phase. Both the film
and the droplets impede the flow of gas through the pipe. It has been reported in [62]
that, liquid loads of 5.6m3/Million m3 of gas reduced the transmission factor of the
pipeline network system by 1%.
Ellul et al. [63] presented the summary of the basic available equations in
multiphase flow conditions. The authors described the current available techniques
for the analysis of multiphase flows. The two methods for the analysis of multiphase
flow conditions are empirical approach and mechanistic approach. The former is
developed based on setting a correlation among parameters of the flow on the basis
of the experimental data for the range of conditions. The later is developed based on
the physical phenomenon of the fluids. One of the limitations that are observed in the
empirical approach is that, the method primarily aims to produce correlation valid
mainly over range of the measured data. On the other hand, the mechanistic
32
modeling approach could be used to generate relationships which are useful for wide
range of data.
Golczynski [64] emphasized on the importance of addressing the effect of
multiphase during the design of pipeline network for optimal operation of TPNS.
Asante [61] proposed the application of various two-phase flow analysis models
based on the liquid holdups. For liquid holdups less than 0.005 (typical in gas
transmission system), homogeneous approach has been recommended. When the
liquid holdups is greater than 0.005 (which is common in gas gathering pipeline
systems), stratified two-phase flow approach has been recommended.
Boriantoro and Adewumi [65] and Leksono [66] developed an integrated
single/two- phase steady state hydrodynamic model for predicting the phase change
and flow regime of fluid flow in pipes. The model predictive capability has been
tested using limited field data and published data in flow regimes. Stanley and
Vaderford [67] presented an online simulation to track the operation of two-phase
wet gas pipelines. The simulation continually receives supervisory control and data
acquisition (SCADA) updates assuring accurate conformance to the actual pipeline.
Taitel and Dukler [68] developed a mechanistic model for analytical prediction
of transition between flow regimes. The approach also provided considerable insight
into the mechanisms of the transitions. The model was tested against data mainly
collected in small diameter pipe under low pressure conditions. An attempt of flow
regime prediction for inclined, large diameter pipes has been done by Wilkens [69]
which could be used as basis for developing the pressure drop equations.
More recently, Shoham [70] developed a mathematical mechanistic model for
predicting of various two-phase flow behaviors. The model is based on the physical
phenomenon of the important flow parameters. However, the derivation of the basic
equations which represents the entire physical phenomenon requires rigorous
computations.
33
Generally, as it is reported in various literatures [24], [66], and [70], single phase
modeling approaches might not be adequate enough to predict the transport
capabilities of the pipelines when gas and liquids move as mixture. A two-phase
analysis or single phase analysis with modified friction factors may be required to
adequately predict the transport capabilities of such system.
There are various two-phase flow models reported in literatures [62], [65], and
[70]. The types of two-phase flow models that could be implemented for the system
depends on the nature and the amount of liquid that exist within the system. The
parameters of two-phase flow like the pressure drop, liquid holdup, and others are
strongly dependent on the existing flow pattern. As a result, the determination of
flow pattern is a key issue in two-phase flow analysis. The following sections discuss
the basics of flow patterns and the detail of homogeneous flow patters which is the
common flow pattern that exists in transmission pipeline network system.
2.4.1.1 Flow Pattern
The review of literatures on flow patterns reveals that there have been variations
among two-phase flow researchers on the definition and classification of flow
patterns. One of the main variation resulted from the complexity of flow phenomena
that occurred during the two-phase flow.
Flow pattern is the geometrical configuration of the gas and liquid phases in the
pipe and it occurs in a gas-liquid two-phase flow. When a gas and a liquid flow
simultaneously in a pipe, the two phases can distribute themselves in a variety of
flow configurations. The flow configurations differ from each other in the special
distribution of the interface, resulting in different flow characteristic, such as velocity
and holdup distributions. Flow pattern in a given two-phase flow system depends on
operational parameters, geometrical variables, and physical properties of the two-
phases [70].
34
Flow pattern developed by Taitel and Dukler [68] is the most commonly used
flow pattern in the analysis of two-phase flow. Figure 2.4 shows the flow patterns
existing in horizontal and near-horizontal (± 100 inclined) pipes developed by
Shoham [70].
Figure 2.4 Flow pattern in horizontal and near-horizontal pipes [70]
The existing flow patterns in horizontal and near-horizontal flow configurations
can be classified as [9] and [70]:
• Stratified smooth flow: this flow pattern occurs at relatively low gas and
liquid flow rates. The two phases separated by gravity, where the liquid-
phase flows at the bottom of the pipe and the gas-phase on the top. The
interface between them is smooth.
• Stratified wavy flow: increasing the gas velocity in a stratified flow, waves
are formed on the interface and travel in the direction of flow.
35
• Elongated bubble flow: the mechanism of the flow in the elongated bubble
flow is that of a fast moving liquid slug overriding the slow-moving liquid
film ahead of it. It occurs when the liquid slug is free of entrained bubbles
and at lower gas flow rates. This flow is sometimes referred to as plug flow.
• Slug flow: occurs at higher gas flow rates, where the flow at the front of the
slug is in the form of an eddy which entrained bubbles.
• Annular flow: this flow occurs at a very high gas flow rate. The gas-phase
flows in a core of high velocity, which may contain entrained liquid droplets
and the liquid flows as a thin film around the pipe wall.
• Wavy annular flow: this flow occurs at the lowest gas flow rates. Most of
the liquid flows at the bottom of the pipe while aerated unstable waves are
swept around the pipe periphery and wet the upper pipe wall occasionally.
This flow occurs on the transition boundary between stratified way, slug and
annular flow.
• Dispersed bubble flow: this flow is one of the homogeneous flows where
either of the two-phases flowing simultaneously in the pipeline is completely
dispersed in the other. Dispersed bubble flow occurs at very high liquid rates.
On the other hand, at a very high gas rates coupled with low liquid loading
the flow is termed as mist flow. In both dispersed bubble flow and mist flow,
the two phases move at the same velocity, and the flow is considered
homogeneous with no-slip.
2.4.1.2 Homogeneous Flow Model
The homogeneous (dispersed bubble and mist) flow models treat the gas-liquid
mixtures as a pseudo single phase with average fluid properties [70] and [71]. It was
reported in [61], low load liquids ( usually less than 0.005) holdup exist in natural
36
gas transmission pipeline network systems. Hence, the best flow pattern to describe
the two-phase gas-liquid mixtures in gas TPNS is mist flow. Because of this,
homogeneous (pseudo single phase) approach is applied in this thesis to describe the
behavior and transport properties of the system of gas and low loads of liquids in the
pipeline.
The model for homogenous two-phase flow analysis is developed based on the
assumption of compressible flow, variable cross-sectional area of pipe, variable
quality of the gas along the length of the pipe [70]. The conservation equations of
mass, momentum, and energy for the model are developed using a control volume
with a cross-sectional are PA and differential length dL in the axial direction as
shown in Figure 2.5.
Figure 2.5 Schematic of homogeneous no-slip flow model adapted from [70]
The continuity equation for the mixture is given by
( )11.2ConstantAvM Pmm == ρ
where M is the total mass flow rate, and mρ and mv are the mixture average
density and velocity, respectively.
A momentum balance on the control volume can be defined as
37
( )12.2sinθρτ gASdLdpA
dLdvM mPwPP
m −−−=
where L is the axial direction, P is the pressure, wτ is the shear stress at the
wall, PS is the pipe perimeter, and θ is the inclination from the horizontal.
Dividing equation (2.12) by the cross-sectional area PA and solving for the
pressure gradient yields
( )13.2sindLA
dvMgAS
dLdp
P
mmw
P
P ++=− θρτ
As shown in equation (2.13), the total pressure gradient equation is composed of
three components: frictional, gravitational, and acceleration components. This
equation will be further developed in Chapter 3 to enable the calculation of each of
the pressure gradient components and the total pressure gradient which provide one
of the governing simulation equations.
2.4.2 Simulation Model with Temperature Variations
One of the variations among the different models developed for the analysis of
TPNS is whether the model takes into account the effect of temperature or not. The
model for the analysis of TPNS can be done on the basis of constant gas flow
temperature i.e. isothermal condition. In reality, as indicated in Menon [23], the
temperature of the gas in TPNS varies along the length of the pipeline due to heat
transfer between the gas and the surrounding soil.
Osiadacz and chaczykowski [72] made a comparison of flow of gas in gunbarrel
pipelines under isothermal and non-isothermal conditions. A significant pressure
profile difference along the pipeline was reported between the isothermal and non-
isothermal analysis. This difference increases with increase in quantity of gas
transmitted. Thermal model for single pipe is also presented in [73].
38
For the variation of gas temperature in TPNS, Menon recommended the
calculation of pressure drop to be done by considering short lengths of pipe that
make up the total pipeline. Figure 2.6 shows a buried pipeline transmitting gas from
node A to node B. Considering a short segment with length of LΔ from the pipeline,
the variation of temperature along the pipeline can be analyzed by applying the
principles of heat transfer.
Figure 2.6 Analysis of temperature variations adapted from [23]
When the gas flows from upstream end with temperature of 1T to downstream
end of the pipe segment with temperature of 2T , the temperature of the gas will drop.
On the other hand, there will be heat transfer between the gas and surrounding due to
temperature difference. The temperature equation relating the two nodes 1 and 2 of
the segment can be expressed as [23]
( )14.2)( 12θ−−+= eTTTT ss
( )15.2PMCLUDΔ
=πθ
where sT is the average soil temperature surrounding pipe segment, M is
mass flow rate of the gas, PC is average specific heat of gas, U is the overall
heat transfer coefficient, and D is the diameter of the pipe.
It can be seen from equation (2.14) that as the pipe length increases, the term θ−e
approaches zero and the temperature 2T becomes equal to soil temperature, sT .
39
Therefore, in a long gas pipeline, the gas temperature ultimately equals the
surrounding soil temperature.
The temperature of the gas is also affected by the compression process. In
adiabatic compression of natural gas, the final temperature of the gas can be
determined knowing the initial temperature and initial and final pressures. From the
adiabatic compression equation and perfect gas law, the discharge temperature is
given by [23, 60]:
( )16.2
1
1
2
2
1
1
2 kk
PP
ZZ
TT
−
⎟⎟⎠
⎞⎜⎜⎝
⎛⎟⎟⎠
⎞⎜⎜⎝
⎛=⎟⎟
⎠
⎞⎜⎜⎝
⎛
where 1Z and 2Z are gas compressibility factor at suction and discharge side
respectively and k is the specific heat ratio.
From equations (2.14) and (2.16), it can be seen that the variations of
temperature in gas TPNS are controlled by these equations. Hence, these equations
will be further developed in Chapter 3 to be incorporated into the governing
simulation equations.
2.4.3 Internal Corrosion
Corrosion in oil and gas industries is one of the serious challenges which affect
the performance of the pipeline network system. As presented in [74], corrosion is
the destructive attack of a metal by chemical or electrochemical reaction with its
environment. The phenomenon of corrosion involves reactions which lead to the
creation of ionic species, by either loss or gain of electrons. Corrosion of material
depends on several factors. Some of them includes: nature of the material or alloy,
surface condition/roughness, composition, moisture absorptivity, environment,
temperature, humidity and corrosive elements.
40
It has been reported in [22] that an increase in ages of the pipes resulted in higher
pipe roughness due to the accumulation of various elements around the internal
surface of the pipe. Figure 2.7 shows the effect of service life of the pipe on
roughness. An increase in roughness has major effect on the performance of the
transmission system i.e. lower flow rate capacity and high pressure drop as indicated
in [23].
Figure 2.7 Effect of years in service on pipe roughness [22]
So far, limited information is available from literature about how the roughness
of the pipes varies with the age of the pipe. However, there are several studies
conducted on corrosion of pipelines for both single phase flow and two-phase gas-
liquid mixtures flow [75]-[78]. One of the issues addressed in this thesis is that, the
effect of ages of the pipes on the performance of TPNS has been incorporated within
the simulation model.
41
2.5 Solution Principles Applied in Gas Pipeline Network Simulations
The selection of an appropriate solution scheme is essential in order to get the
required parameters which are necessary to analyze the performance of the TPNS.
The most commonly used solution schemes for analyzing the governing equations of
the simulation are successive substitution (SS) and Newton-Raphson (NR) solution
schemes[79].
Successive substitution scheme [79] and [80] has been applied for simulation of
water pumping system [81] and for obtaining solutions for boundary value problems
[82]. The major advantage of using successive substitution scheme is that the method
is easy to use when the numbers of equations within the system are few.
Newton-Raphson scheme [79]-[81] has been also applied in gas pipeline network
simulation and other networks for various analysis [83]-[86]. One of the major
advantages of Newton-Raphson algorithm is that the method is more reliable and
rapidly convergent. Furthermore, it is not necessary to list the equations in any
special order. Abbaspour et al [83] used the Newton-Raphson algorithm to solve the
nonlinear finite difference thermo-fluid equations for two-phase flow in a pipe. Brkic
[84] used Newton-Raphson method to solve the nonlinear flow equations in natural
gas distribution networks. Beck and Boucher [85] also applied the Newton-Raphson
algorithm to analyze steady state fluid circuits. Kessal [86] used the Newton-
Raphson algorithm for analyzing fluid flow in gas pipelines.
The maximum relative percentage error at each iteration for both successive
substitution and Newton-Raphson is calculated based on the relationships given as
[79]:
( )17.2100)]max(/)([max% ×−= newoldnew XXXErrorageMaximum where X is the vector that represents the unknown variables.
42
In order to take the advantages of both solution schemes, two solutions are
developed in this thesis. The first one is based on successive substitutions and
applied for simple TPNS configurations. The second one is a generalized solution
scheme based on Newton-Raphson which is applied for complex TPNS
configurations. Detailed analysis of the solution schemes for the application of TPNS
simulation is discussed in Chapter 3.
2.6 Summary
This Chapter presented the review of optimization of gas pipeline network
systems, simulation of gas pipeline networks, the simulation models with two-phase
flow, temperature variations and corrosion, and solution principles applied in gas
pipeline network simulation.
The various TPNS simulation models revised in this thesis are summarized in
Table 2.2. From the review of the TPNS simulation models, it is observed that while
there is significant progress on the area of TPNS simulation methodologies, there are
still issues that need to be addressed. The effects of the compressor characteristics,
two-phase flow and internal corrosion during TPNS simulation modeling are
addressed in this research.
43
Table 2.2 Summary of TPNS Simulation Models
Method Author Main features Limitations
Newton-loop Osiadacz [7] • Preset the discharge pressures at each compressor stations.
• Single phase flow and constant temperature
• Speed, discharge pressures, suction temperature neglected
• Constant corrosion Black box Letniowski [21] • Consider the compressor station as a black
box • Single phase flow, constant temperature
• Speed, discharge pressures, suction temperature neglected
• Constant corrosion Graph theory Wu et al. [47] • Describing the feasible region of
compressors mathematically and optimizing the operation of TPNS.
• Single phase flow , constant temperature
• Based on approximation of fuel cost functions.
• Constant corrosion
Correlation Nimmanonda
et al [16] and [18]
• Continuity, mass balance, and energy balance equations used.
• The relationship between compression ratio, BHP, and flow was developed for three flow categories.
• Single phase flow, constant temperature
• Compressor stations handling and only TPNS without loop
• Requires presetting either the suction or discharge pressure
• Speed of the compressors was not taken into considerations.
Mathematical approximation
Abbaspour [9] • Use of polynomial approximation for modeling the compressor to develop simulation based optimization.
• Two phase flow and variable temperature
• Perform optimization or simulation on station level.
• Constant corrosion
44
Most of the studies on optimization of pipeline network system used either
relaxation of the constraints or approximation of the constraints with equivalent
constraints to facilitate for computation. These could result in deviation from the
original problem. Furthermore, the optimization of pipeline gives only the optimal
operation of the system. On the other hand, simulation of the pipeline network
system is used to investigate the off design operation of the system.
The review of simulation models developed for natural gas TPNS indicated that
the models were either limited for pipe only or treat the non-pipe elements based on
simplified approach by neglecting the importance of the parameters such as speed,
suction pressure, discharge pressure, and suction temperatures. In some cases, the
applications of the models may be limited to either simple pipeline configurations or
fixed environmental conditions. Pressure drop and flow rate of the gas may be
affected as the age of the pipe increase. However, limited studies on the relationships
between the age of the pipe and its effect on pressure drop and flow are reported on
literatures. The study conducted on non-isothermal model of TPNS lies mostly on
single pipe or limited to simplified network configurations like gunbarrel.
In the following Chapter, the details of the development of the methodology to
address the objective of this research are presented. The basic principles adopted in
this Chapter are utilized to develop the methodology.
45
CHAPTER 3
METHODOLOGY
3.1 Introduction
In this Chapter, a suitable methodology was developed to achieve a TPNS
simulation model for performance analysis of gas pipeline networks. The
determination of operational and design variables for the TPNS were conducted
based on the analysis of factors involved during the transmission process.
Appropriate mathematical formulations and suitable solution procedures were
established to get reliable simulation results that predict the actual situations. The
basic equations for the TPNS simulation were derived from the principles of flow of
fluid through pipe, compressor characteristics and the principles of mass balance at
the junction of the network.
First, the mathematical formulation of the TPNS simulation model is discussed.
The basic single phase flow equations, compressor characteristics equations, mass
balance and looping conditions which form the governing simulation equations are
presented. Then, the solution schemes to get the flow and pressure variables based on
iterative successive substitution and Newton-Raphson schemes are presented. The
enhanced Newton-Raphson based TPNS simulation model which contains additional
features such as two-phase flow analysis, temperature variations and internal
corrosion is also discussed. The last two sections present the performance evaluation
of TPNS and summary of the Chapter. Figure 3.1 shows the structure of the
simulation model for natural gas TPNS.
47
3.2 Mathematical Formulation of the Simulation Model
The mathematical formulation and the types of equations incorporated into the
governing simulation equations depend on the configurations of the network and
elements of the TPNS as shown in Figure 3.1. The mathematical model for the TPNS
simulation is developed based on equations which govern the flow of the gas through
pipes, the performance characteristics of the compressors, and the principles of
conservation of mass.
3.2.1 Single Phase Flow Equations Modeling
One of the governing equations for the simulation is derived based on the
principle of flow analysis of gas in pipes. The flow of gas through pipes can be
affected by various factors such as the gas properties, friction factor, and the
geometry of the pipes. As discussed in section 2.3.1, several flow equations are in
use in the gas industry for single phase gas flow. In general, for a horizontal pipe
connecting node i and j (Figure 3.2), the general flow equation for relating upstream
pressure iP , downstream pressure jP and the flow through pipe ijQ can be expressed
as:
( )1.3222ijijji QKPP =−
where ijK is to be determined by the gas properties and the characteristics of
pipe connecting node i and j .
48
Figure 3.2 Pipe joining two consecutive nodes
Equation (3.1) can be represented as functional form [81]. If all nodal pressures
and flow rate are unknown, the functional representation of equation (3.1) takes the
form:
( )2.30),,( =ijji QPPf
If the upstream pressure and the flow rate are the only unknowns, the functional
representation for equation (3.1) takes the form
( )3.30),( =iji QPf
Note that functional form of representation for an equation consists of only
parameters which are unknown in that equation. Figure 3.3 shows part of the TPNS
consisting of three compressor stations (CS1, CS2 and CS3) and six pipes i.e. pipe 0-
1, 2-3, 3-4, 5-6, 3-7 and 8-9.
49
Figure 3.3 Part of transmission pipeline network system
Assuming that the source pressure at node 0 and the demand pressures at node 6
and 9 are known, the general flow equations for the pipes for single phase flow and
the corresponding functional representation can be summarized as shown in
Table 3.1.
Table 3.1 Single phase flow equations and functional representations
Pipe node General flow equation Functional representation
Start node End node
0 1 2101
21
20 QKPP =− 0),( 111 =QPf
2 3 2123
23
22 QKPP =− 0),,( 1322 =QPPf
3 4 2234
24
23 QKPP =− 0),,( 2433 =QPPf
3 7 2337
27
23 QKPP =− 0),,( 3734 =QPPf
5 6 2256
26
25 QKPP =− 0),( 255 =QPf
8 9 2389
29
28 QKPP =− 0),( 386 =QPf
For ease of representation, from now onwards, the term flow equation will be
frequently used to mean the general flow equation. Furthermore, the functional form
of representation will be used interchangeably with the flow equation.
50
3.2.2 The Looping Conditions
When the TPNS contains loops, additional equations must be incorporated to the
flow equations for the pipes. These additional equations are obtained from looping
condition. The looping condition states that for each closed loop within the network
system the pressure drop is zero [7, 23].
Looped piping system, as shown in Figure 3.4, consists of two or more pipes
connected in such a way that the gas flow splits among the branch pipes and
eventually combine downstream into a single pipe. It can be constructed of the same
diameter pipe as the main pipeline or based on different size. The reason for
installing loops is to reduce pressure drop in a certain section of the pipeline due to
pressure limitation or for increasing the flow rate in bottleneck sections. For instance
for the TPNS shown in Figure 3.4, by installing a pipe loop from node 3 to node 6,
the overall pressure drop can be reduced due to the split of flow rate through the two
pipes. Based on the looping condition, the pressure drop in pipe branch 3-C1-4 must
equal the pressure drop in pipe branch 3-C2-4. This is due to the fact that both pipe
branches have a common starting point (node 3) and common ending point (node 4).
51
Figure 3.4 Looped pipeline network system
The pressure drop due to friction for single phase flow in branch 3-C1-4 can be
formulated based on the general flow equation as
( )4.351
22112
62
3 DQLKPP =−
where, 1K is a parameter that depends on gas properties and friction factor,
2Q is the flow rate, 1L and 1D are the length and diameter of the pipe branch
3-C1-4, respectively.
Similarly, the pressure drop due to friction for single phase flow in branch 3-C2-4
can be given as
( )5.352
23222
62
3 DQLKPP =−
where, 2K is a parameter that depends on gas properties and friction factor,
3Q is the flow rate, 2L and 2D are the length and diameter of the pipe
branch 3-C2-4, respectively.
52
In equations (3.4) and (3.5), the constants 1K and 2K are equal, since the same
gas is flowing through both branch pipes. Combining both equation results in
( )6.352
232
51
221
DQL
DQL
=
Further simplification of equation (3.6) gives
( )7.35.2
2
15.0
1
2
3
2⎟⎟⎠
⎞⎜⎜⎝
⎛⎟⎟⎠
⎞⎜⎜⎝
⎛=
DD
LL
Equation (3.7) is referred to as the looping condition and it is one of the
governing simulation equations when the pipeline network system contains loops.
For the unknown flow rate 2Q and 3Q , the functional representation for equation
(3.7) takes the form
( )8.30),( 32 =QQf
3.2.3 Compressor Characteristics Equations
When the TPNS contains non-pipe elements, one of the governing simulation
equations should have to be derived from compressor. Compressors are usually
characterized by the performance map. In order to integrate the performance
characteristics of compressor to the governing simulation equations, each of the
constant speed curves from the map should have to be represented by mathematical
model. Even though this type of representation results an accurate approximation for
the curves, it is time consuming. Hence, the compressor map is usually represented
based on single curve by using normalized parameters.
53
Figure 3.5 shows the plot of normalized head against the normalized flow rate for
typical centrifugal compressor data taken from [55] for speeds of 8856, 8000, 7000,
6000, 5000, 4000, and 3630rpm. The corresponding plot for efficiency against the
normalized flow rate is shown in Figure 3.6. It can be seen from both the figures that
all the constant speed curves had the tendency to coincide as single line.
Figure 3.5 Normalized head and normalized flow rate for different speeds of a
typical centrifugal compressor
54
Figure 3.6 Efficiency and normalized flow rate for different speeds of a typical
centrifugal compressor
In considering the effect of compressors for the TPNS simulation model, the
relationships as in equation (2.9) and equation (2.10) might not be used directly. The
information from the compressor map should have to relate the discharge pressure,
the suction pressure and flow rate. The relationships between suction pressure sP ,
and discharge pressure dP with the head H is given as [60] :
( )9.31⎪⎭
⎪⎬⎫
⎪⎩
⎪⎨⎧
−⎥⎦
⎤⎢⎣
⎡=
m
s
dS
PP
mZRT
H
where ( ) kkm /1−= with k to be specific heat ratio, R is gas constant, ST is
the suction temperature and Z is the compressibility of the gas.
55
Substituting the value of H from equation (3.9) into equation (2.9) and
rearranging yields the required compressor performance equation which can be
incorporated as governing equation for the simulation model.
( ) ( ) ( ){ } ( )10.31/// 34
2321
2++++=⎟⎟
⎠
⎞⎜⎜⎝
⎛nQAnQAnQAA
ZRTmn
PP
S
m
s
d
Equation (3.10) represents a general compressor equation for single compressor.
It can be seen that most of the parameters that describe the compressor are
incorporated in the general compressor equation. This is one of the significant
contributions in the area of simulation of TPNS with compressor stations as non-pipe
elements.
If the suction side pressure, discharge side pressures and flow rates are unknown,
equation (3.10) can also be represented with short functional form as
( ) ( )11.30,, =QPPf sd
The coefficients 4321 A,A,A,A used in equation (2.9) and the corresponding
coefficients 4321 B,B,B,B for equation (2.10) are unique and depend on the units used
in the compressor map. For gas pipeline compressors taken from [55] and [59], the
different coefficients for equation (2.9) and (2.10) are summarized as shown in
Table 3.2.
56
Table 3.2 Values of coefficients for compressor equations
Types of compressor Coefficients for
equation (2.9)
Coefficients for
equation (2.10)
Centrifugal compressor
from [55]
A1 1.20 x 10-06 B1 9.7 x 10-01
A2 -2.48 x 10-09 B2 -1.14 x 10-02
A3 -4.60 x 10-12 B3 2.56 x 10-04
A4 -3.17 x 10-13 B4 -1.51 x 10-06
Centrifugal compressor
from [59]
A1 1.08 x 10-06 B1 7.99 x 10-01
A2 -1.96 x 10-08 B2 -1.03 x 10-02
A3 4.71 x 10-10 B3 6.19 x 10-04
A4 -5.83 x 10-12 B4 -7.86 x 10-06
Equation (3.10) represents a general compressor equation for single compressor.
However, there are cases where several compressors may work within compressor
stations either on the basis of serial or parallel arrangements. When the compressors
operate in series within the station as shown in Figure 3.7, compressor equations for
each compressor should have to be developed to represent the effect of all
compressors.
Figure 3.7 Compressors operating in series within a station
For instance, if there are cn number of compressors operating in series within the
station, there will be cn number of independent compressor equations for the station.
57
Equation (3.12) shows the compressor equation for the first compressor within the
station.
( ) ( ) ( ){ } ( )12.313/42/3/21
2
1
1 ++++=⎥⎦
⎤⎢⎣
⎡nQAnQAnQAA
sZRTnm
PP
m
s
d
where 1sP and 1dP are the suction and discharge pressures for the first
compressor.
The remaining compressor equations within the station were developed by
following the same procedure as in equation (3.12).
For compressors operating in parallel within the stations as shown in Figure 3.8,
only a single equation may represent the compressor station assuming that identical
compressors are working within the station. For instance, for the compressor station
with cn number of compressors working in parallel, the general compressor equation
can be modified to represent the compressor station as
[ ] [ ] [ ]{ } ( )13.313)/(42)/(3)/(21
2+++=⎥
⎦
⎤⎢⎣
⎡ncnQAncnQAncnQAA
ZRTnm
PP
S
m
s
d
where sP and dP are the suction and discharge pressures for the compressor
station.
58
Figure 3.8 Compressors working in parallel within a station
The general compressor equation was validated based on the performance map
of the compressor. Figure 3.9 shows a comparison of the plot of the performance
characteristics of the compressor generated by Equation 3.10 and actual data
collected from the performance map of the compressor shown in Figure 2.3. It is
observed that the maximum percentage error introduced when the performance of
the map is approximated by the general compressor equation was 3%.
59
Figure 3.9 Comparison of selected data and approximated data based on the
compressor data from [59]
Figure 3.10 shows the plot of the approximated pressure ratio ( )PsPd / based on
Equation 3.10 against actual pressure ratio from the characteristic curve for the
typical compressor taken from [55] at speed of 8000 rpm. As the speed of the
compressor increased, the deviation between the actual and calculated pressure ratio
increased which made the approximation a bit far from the real values. For gas
pipeline centrifugal compressor where the maximum pressure ratio is limited to
around 1.5, the mathematical model developed in Equation 3.10 could be taken as
reasonable approximation for the performance characteristics.
60
Figure 3.10 Comparison of calculated and actual pressure ratios
3.2.4 Mass Balance Equations
In addition to the pipe flow equations, looping conditions, and compressor
equations, mass balances provide the remaining basic equations for the simulation of
TPNS configurations with branches. The mass balance equations were obtained
based the principle of conservation of mass at each junction of TPNS.
At any junction node c within the TPNS, shown in Figure 3.11, the generalized
mass balance equation for t incoming pipes, u outgoing pipes, and w load pipes can
be summarized as:
( )14.3011 1
=−− ∑∑ ∑=
=
=
=
=
=
wl
ll
ti
i
uj
jji DqQ
where tQQQ ,....,, 21 are flow through incoming pipes to junction c , uqqq ,....,, 21 are flow through outgoing pipes from junction c and lDDD ,....,, 21 are the
load from junction node c .
61
Figure 3.11 Mass balance formulation
The functional representation of equation (3.14) takes the form shown in
equation (3.15) if all the flow rates through the incoming and outgoing pipes are
unknown.
( )15.30)...,,,,...,,,( 2121 =ut qqqQQQf
3.3 Successive Substitution based TPNS Simulation Model
In section 3.2, the details of the mathematical formulation for a given TPNS have
been presented. The number of variables (flow and pressure) to be determined
depends on the configurations (number of pipes, number of compressor stations,
number of branches, and number of loops). The principles of flow through pipes,
performance characteristics of the compressor, conservation of mass, and looping
conditions were able to provide sufficient number of equations to determine the
unknown variables.
Once the basic governing simulation equations for TPNS are developed, the
result for unknown pressure and flow variables could be determined. The most
commonly used solution schemes for analyzing the governing equations of the
simulation are iterative successive substitution and Newton-Raphson solution
schemes. The major advantage of using successive substitution scheme is that the
method is easy to use when the numbers of equations within the system are few. One
62
of the advantages of Newton-Raphson algorithm is that the method is more reliable
and rapidly convergent. Furthermore, it is not necessary to list the equations in any
special order.
In order to take the advantages of both solution schemes, two solutions are
developed in this thesis. The first one is based on successive substitutions and
applied for simple TPNS configurations. The second one is a generalized solution
scheme based on Newton-Raphson which is applied for complex TPNS
configurations.
A relationships between the number of pipes np, number of compressor stations
ns with compressors working in parallel, number of loops nl, and the number of
branches (junctions) nj with the total number of unknown pressure and flow
variables are developed. Table 3.3 shows the summary of the relationships between
the number of equations available and the total number of unknown variables.
Table 3.3 Number of equations and unknowns for TPNS configurations
Items No. of equations No. of unknowns
Flow
Comp. Mass balance
Looping conditions
P
Q
Basic configurations and elements of TPNS
Pipes np - - - (np+ns) – (nj+1)
2nl + 2nj + 1
Comp. - ns - - Branches - - nj - Loops - - - 2nl
Total number of equations and unknowns
np + ns + nj +2nl
np + ns + nj +2nl
The data shown in Table 3.4 were used throughout the analysis for the numerical
evaluations. All the properties of the gas were collected from the nearest operational
gas transmission company. For the compressor station, all units within the stations
are assumed to be identical and are connected in parallel. The data related to pipe
diameters, lengths and customer requirements are based on existing TPNS and
literatures. The dimensions used for the parameters are ][kPaP , ][KT , ][kmL , ]/[ 3 hrmQ
and ][mmD . The pipe flow resistance 50960.7 DfLEKij ×+= .
63
Table 3.4 Gas properties
In this section, successive substitution based TPNS simulation model is
discussed. The application of the model is demonstrated based on two pipeline
network configurations.
When the successive substitution scheme is to be applied for the determination of
unknown variables for the given TPNS, information flow diagram (IFD) is an
important tool [81]. IFD involves the representation of the basic equations of the
TPNS as an input-output information blocks and arranging them in such a way that
only one output can be calculated from each block. A compressor might appear in
TPNS diagram like the one shown in Figure 3.3. But, the information flow block of
the compressor might take one of the forms shown in Figure 3.12. Note that the
blocks in this figure represent a compressor equation presented in function form as
( ) 0,, 21 =PPQf . In order to develop the IFD for the TPNS, all the components of the
network should have to be represented as information block and arranged in such a
way that only one output can be determined from single block.
Gas properties Numerical values used for analysis
Gas composition
Methane 92%
Ethane 5%
Nitrogen 1%
Others 2%
Gas gravity (G) 0.5
Gas flowing temperature (T) 308 K
Base pressure (Pn) 101kPa
Base temperature (Tn) 288K
Gas constant for air (Rair) 287.5J/kgK
Gas compressibility (Z) 0.91
Isentropic exponents (k) 1.287
64
Figure 3.12 Possible information flow blocks for compressor
The method of successive substitution is closely associated with the IFD of the
system. The solution procedure starts by assuming a value of one or more variables,
beginning the calculation, and proceeding through the system until the originally
assumed variables have been recalculated. The recalculated value is then substituted
successively and the calculation loop is repeated until satisfactory convergence is
achieved. Figure 3.13 shows the developed flowchart of the simulation model based
on the iterative successive substitution scheme.
66
3.3.1 Case 1A: Single Compressor Station and Two Customers Module
In single compressor station two customers network, it is required to transmit gas
from source to two different customers with the required pressure and flow rate.
Figure 3.14 shows schematic diagram of the TPNS when the gas is delivered from
source to two different customer sites of station D1 and D2 using single compressor
station (CS).
Figure 3.14 Gas transmission system for two customers
The TPNS consists of 4 pipes, 1 compressor station, 1 junction, and with no loop.
Therefore, np = 4, ns = 1, nl = 0, and nj = 1. As a result, based on Table 3.3, there are
3 nodal pressures and 3 flow variables to be determined. A total of 6 independent
equations were obtained in order to solve the network problem. The basic governing
equations for TPNS were developed from pipe flow equations, compressor stations
equations and mass balance equations based on the discussion in section 3.2. There
are 4 pipe flow equations, 1 compressor station equation and 1 mass balance
equation which form the required number of equations to solve for the unknown
variables.
The summary of flow equations and their corresponding functional
representation for the given TPNS are shown in Table 3.5.
67
Table 3.5 Summary of flow equations and their corresponding functional
representations for the given TPNS
Pipe node Flow equation Functional representation
Start node End node
0 1 2101
21
20 QKPP =− 0),( 111 =QPf
2 3 2123
23
22 QKPP =− 0),,( 1322 =QPPf
3 D1 2113
21
23 CDD QKPP =− 0),( 133 =CQPf
3 D2 2223
22
23 CDD QKPP =− 0),( 234 =CQPf
The compressor equation for the TPNS is given as:
( ) ( ) ( ) ( )16.31]///[ 314
213121
2
1
2 ++++=⎟⎟⎠
⎞⎜⎜⎝
⎛nQAnQAnQAA
ZRTmn
PP
S
m
The corresponding functional representation for equation (3.16) takes the form
( ) ( )17.30,, 1215 =QPPf
The mass balance equation is given by:
( )18.3211 CC QQQ +=
The corresponding functional representation for equation (3.18) takes the form
( ) ( )19.30,, 2116 =CC QQQf
Based on the functional representations for the mathematical formulation of the
governing simulation equations for the given TPNS, the IFD can be drawn as shown
in Figure 3.15. Using this IFD, the successive substitution solution scheme could be
started by assuming one or more variables. For the given TPNS, the solutions can be
obtained by assuming initial value for 1Q and 3P . Once these variables are assumed,
the value of 2CQ can be calculated, then 1CQ , 2P , etc. by following the IFD. The
68
iteration will continue till the desired error limits or the number of iterations is
achieved.
Figure 3.15 IFD of pipeline network system with two customers
The application of the SS based TPNS simulation model for the given network
was conducted based on nodal pressure requirements and pipe data shown in Table
3.6 and Table 3.7 , respectively.
Table 3.6 Pressure data for single CS and two customers
Node Pressure[kPa]
D1 1800
D2 1800
Source 4000
Table 3.7 Pipe data for single CS and two customers
Pipe node Diameter [mm] Length [km]
Start node End node
0 1 400 80
2 3 400 60
3 D1 400 80
3 D2 300 60
69
3.3.2 Case 2A: Two Compressor Stations and Two Customers Module
The problem in two compressor stations and two customer network is to transmit
gas from source to two customers using two compressor stations. Figure 3.16 shows
TPNS when gas is delivered from source to two different customers’ sites designated
as 1D and 2D using two compressor stations.
Figure 3.16 Pipeline network with two CSs and two customers
This TPNS consists of 5 pipes, 2 compressor stations, 1 junction, and with no
loop. Therefore, np = 5, ns = 2, nl = 0, and nj = 1. As a result, based on Table 3.3,
there are 5 nodal pressures and 3 flow parameters to be determined. Therefore a total
of 8 independent equations should have to be obtained in order to solve the network
problem. The basic governing equations for TPNS were developed from pipe flow
equations, compressor stations equations and mass balance equations based on the
discussion in section 3.2. There are 5 pipe flow equations, 2 compressor station
equations and 1 mass balance equation which formed the required number of
equations to solve for the unknown parameters.
The summary of flow equations and their corresponding functional
representation for the given TPNS are shown in Table 3.8.
70
Table 3.8. Summary of flow equations and their corresponding functional
representations for the given TPNS
Pipe node Flow equation Functional representation
Start node End node
0 1 2101
21
20 QKPP =− 0),( 111 =QPf
2 3 2123
23
22 QKPP =− 0),,( 1322 =QPPf
4 5 2145
25
24 QKPP =− 0),,( 1543 =QPPf
5 D1 2115
21
25 CDD QKPP =− 0),( 154 =CQPf
5 D2 2225
22
25 CDD QKPP =− 0),( 255 =CQPf
The compressor equations for the given TPNS were summarized in equation and
function forms as shown in Table 3.9.
Table 3.9 Summary of compressor equations and functional representations
Compressor
stations
Compressor equation Functional representation
s
CS1 ( ) ( ) ( ) 1]///[ 3114
21131121
21
1
2 ++++=⎟⎟⎠
⎞⎜⎜⎝
⎛nQAnQAnQAA
ZRTmn
PP
S
m
0),,( 1216 =QPPf
CS2 ( ) ( ) ( ) 1]///[ 3214
22132121
22
3
4 ++++=⎟⎟⎠
⎞⎜⎜⎝
⎛nQAnQAnQAA
ZRTmn
PP
S
m
0),,( 1437 =QPPf
Mass balance equation at junction node 5 is given by:
( )20.3211 CC QQQ +=
The corresponding functional representation for equation (3.20) takes the form
( ) ( )21.30,, 2118 =CC QQQf
71
Based on the functional representations for the mathematical formulation of the
governing simulation equations for the given TPNS, the IFD is as shown in
Figure 3.17. Similar procedure was followed as in the case of single compressor
station with two customers. Using this IFD, the successive substitution solution
scheme can be started by assuming two variables. For the given TPNS, the solution
is obtained by assuming initial value for 1Q and 5P . Based on the assumed variables,
the value of 2CQ will be calculated, then 1CQ , 4P , etc. by following the IFD. The
iterations will be continued till the desired error limits or the number of iteration is
achieved.
Figure 3.17 IFD for pipeline network with two CSs
The application of the SS based TPNS simulation model for the given network
was conducted based on nodal pressure requirements and pipe data shown in Table
3.6 and Table 3.10 , respectively.
72
Table 3.10 Pipe data for two CSs and two customers
Pipe node Diameter [mm]
Length [km] Start node End node
0 1 200 100 2 3 200 50 4 5 200 100 5 D1 200 150 5 D2 200 100
3.4 Newton-Raphson based TPNS Simulation Model
As presented in section 3.3, the solution to the unknown pressure and flow
variables are also obtained based on Newton-Raphson solution schemes. Newton-
Raphson technique is complex but powerful for analysis of pipeline network
problems with large number of pipes and compressor stations. The multivariable
Newton-Raphson method which is very important for the analysis of the TPNS can
also be derived following the same procedure as that of the single variable.
A set of nonlinear equations which are obtained from single phase flow
modeling, compressor modeling, looping condition, and mass balance formulations
discussed in section 3.2 can be represented in matrix form.
Let =PN total number of unknown pressure variables.
=QN total number of unknown flow variables.
The total number of unknown variables TotalN is given as:
( )22.3QPTotal NNN +=
The set of single phase pipe flow, looping, compressor, and mass balance
equations can be represented as
73
( )( )
( )
( )23.3
0,,,,,
0,,,,,
0,,,,,
,21,21
,21,212
,21,211
⎪⎪⎪⎪
⎭
⎪⎪⎪⎪
⎬
⎫
⎪⎪⎪⎪
⎩
⎪⎪⎪⎪
⎨
⎧
=
=
=
••••••
•••
•••
•••
••••••
••••••
QNPNTotalN
QNPN
QNPN
QQQPPPF
QQQPPPF
QQQPPPF
Equation (3.23) can be written in matrix form as [79]
( )24.30~~~
=⎟⎟⎠
⎞⎜⎜⎝
⎛ XF
where the vector ~
X represents the total number of unknown pressure, flow
and temperature variables and ~
F is the corresponding equations generated
from pipe, compressor, mass balance and looping conditions.
The multivariable Newton-Raphson iterative procedure for equation (3.24) takes
the form
( )25.3~~
1
`~
~~~
⎟⎟⎠
⎞⎜⎜⎝
⎛
⎥⎥⎦
⎤
⎢⎢⎣
⎡−=
−
old
oldXoldnew XFAXX
where ~A is called Jacobian matrix whose elements are partial derivatives of
the functions with respect to the variables.
The matrix ~A in equation (3.25) is defined as
74
( )26.3
11
2
1
22
1
2
1
1
11
1
1
~
⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥
⎦
⎤
⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢
⎣
⎡
∂∂
∂∂
∂∂
∂
∂
•
∂∂
∂∂
∂∂
∂∂
∂∂
∂∂
∂∂
∂∂
=
••••••
•
•
••••••
••••••
QN
NTotalNTotal
PN
NTotalTotalN
QNPN
QNPN
QF
QF
PF
PF
QF
QF
PF
PF
QF
QF
PF
PF
A
From equation (3.25), the inverse of the Jacobian matrix needs to be computed at
each iterations. However, there is another approach that does not require the rigorous
computation of associated with the inversion of the Jacobian matrix. Equation (3.25)
can be rewritten as[79]
( )27.3~~~~
~
~
⎟⎟⎠
⎞⎜⎜⎝
⎛−=⎥
⎦
⎤⎢⎣
⎡− oldoldnew
oldX
XFXXA
The value of the unknown variables were calculated from equation (3.27)
iteratively until the maximum relative percentage error defined in equation (2.17) is
less than the specified tolerance or the number of iterations reached the desired
value. Figure 3.18 shows the flowchart of the TPNS simulation model based on the
iterative Newton-Raphson solution scheme.
76
A visual C++ code is developed for the TPNS simulation based on Newton-
Raphson solution technique. In order to handle the visual C++ program efficiently,
the overall code is grouped into several subtasks. These include, subtask for
mathematical formulation, subtasks for matrix elements generation, subtask for input
data, subtask for Gaussian eliminations, subtask for error sorting, subtask for
evaluating the networks, etc. The snapshot of the typical TPNS source code for
analyzing part of Malaysian gas transmission system is shown in Appendix A.
Sample simulation results from the Newton-Raphson based TPNS simulation model
for gunbarrel pipeline network is as shown in Appendix B.
The applications of Newton-Raphson based TPNS simulation model is
demonstrated using three different network configurations. These include gunbarrel,
branched and looped TPNS configurations. These network configurations are the
most widely used in natural gas industry. The application of the Newton-Raphson
based TPNS simulation model also demonstrated using part of the Malaysian gas
transmission network system. The applications of the Newton-Raphson based TPNS
simulations are presented in the following sections.
3.4.1 Case 1B: Gunbarrel Pipeline Network Module
In gunbarrel (linear) pipeline network system where the gas is transported from
source to demand station linearly, there are no junctions and loops involved. As a
result, the basic simulation equations include only flow and compressor equations.
Figure 3.19 shows a typical gunbarrel transmission network system with two
compressor stations. This network was used in order to demonstrate the application
of the TPNS simulation model for the gunbarrel pipeline network system.
77
Figure 3.19 Gunbarrel pipeline network system with two CSs
For the TPNS shown in Figure 3.19, the number of pipes np = 3 and the number
of compressor stations ns = 2. As a result, based on Table 3.3, there will be 4 pressure
variables and 1 flow variable to be determined. There are 3 flow equations and 2
compressor equations available to solve the problem.
From the discussion in section 3.2.1, the pipe flow equations for single phase gas
flow based on general flow equation is summarized as shown in Table 3.11.
Table 3.11 Summary of flow equations for the given gunbarrel TPNS
Pipe node Flow equation Functional representation
Start node End node
0 1 201
21
20 QKPP =− 0),( 11 =QPf
2 3 223
23
22 QKPP =− 0),,( 322 =QPPf
4 5 245
25
24 QKPP =− 0),,( 543 =QPPf
Similarly, based on the discussion on compressor stations modeling in section
3.2.3, the remaining compressor equations for the network can be formulated and
represented in functional forms as shown in Table 3.12.
78
Table 3.12 Summary of compressor equations and functional representations
Compressor
stations
Compressor equation Functional representations
CS1 ( ) ( ) ( ) 1]///[ 314
213121
21
1
2 ++++=⎟⎟⎠
⎞⎜⎜⎝
⎛nQAnQAnQAA
ZRTmn
PP
S
m
0),,( 214 =QPPf
CS2 ( ) ( ) ( ) 1]///[ 324
223221
22
3
4 ++++=⎟⎟⎠
⎞⎜⎜⎝
⎛nQAnQAnQAA
ZRTmn
PP
S
m
0),,( 435 =QPPf
Based on the flow equations shown in Table 3.11 and the compressor equations
shown in Table 3.12, the solution to the unknown pressure and flow variables are
obtained using the developed Newton-Raphson based TPNS simulation model.
The application of the Newton-Raphson based TPNS simulation model for the
gunbarrel network was conducted based on nodal pressure requirements and pipe
data shown in Table 3.13 and Table 3.14 , respectively. The pipe data was based on
[47]. The results of the pressure and flow variables and their convergence graphs will
be discussed in Chapter 4.
Table 3.13 Pressure requirements for two CSs and single customer
Node Pressure [kPa]
0 3000
5 4000
Table 3.14 Pipe data for two CSs and single customer
Pipe node
Diameter [mm]
Length [km] Start node End node
0 1 900 80
2 3 900 80
4 5 900 80
79
3.4.2 Case 2B: Branched Pipeline Network Module
The second type of transmission network analyzed using the developed Newton-
Raphson based TPNS simulation model is branched system. In this system, gas is
transported from the source to different demand stations by branching from the main
source. Figure 3.20 shows typical branched transmission network system with one
compressor station and branching to five customers. In order to demonstrate the
application of the simulation model for the branched pipeline network system, this
network is considered for the analysis.
Figure 3.20 Branched pipeline network system
For this TPNS, the number of pipes np = 10, the number of compressor stations
ns = 1, and the number of junctions nj = 4. As a result, based on Table 3.3, there will
be 6 pressure variables and 9 flow variable to be determined. There are ten flow
equations, one compressor equation and four mass balance equations available to
solve the problem. From the discussion in section 3.2.1, the pipe flow equations for
single phase gas flow based on general flow equation are summarized as shown in
Table 3.15.
80
Table 3.15 Summary of flow equations and their functional representation
Pipe node Flow equation Functional representation Start node End node
0 1 2101
21
20 QKPP =− 0),( 111 =QPf
2 3 2123
23
22 QKPP =− 0),,( 1322 =QPPf
3 D1 2113
21
23 CDD QKPP =− 0),( 133 =CQPf
3 4 2234
24
23 QKPP =− 0),,( 2434 =QPPf
4 D2 2224
22
24 CDD QKPP =− 0),( 245 =CQPf
4 5 2345
25
24 QKPP =− 0),,( 3546 =QPPf
5 D3 2335
23
25 CDD QKPP =− 0),( 357 =CQPf
5 6 2456
26
25 QKPP =− 0),,( 4658 =QPPf
6 D4 2446
24
26 CDD QKPP =− 0),( 469 =CQPf
6 D5 2556
25
26 CDD QKPP =− 0),( 5610 =CQPf
The compressor equation for the TPNS is given as:
( ) ( ) ( ) ( )28.31]///[ 314
213121
2
1
2 ++++=⎟⎟⎠
⎞⎜⎜⎝
⎛nQAnQAnQAA
ZRTmn
PP
S
m
The corresponding functional representation for equation (3.28) takes the form
( ) ( )29.30,, 12111 =QPPf
The remaining mass balance equations for the network were formulated and
summarized as shown in Table 3.16.
Table 3.16 Summary of mass balance equations and functional representations
Node Mass balance equation Functional representation
3 121 CQQQ += ( ) 0,, 12112 =CQQQf
4 232 CQQQ += ( ) 0,, 23213 =CQQQf
5 343 CQQQ += ( ) 0,, 34314 =CQQQf
6 544 CC QQQ += ( ) 0,, 54415 =CC QQQf
81
Thus, the solutions to the unknown pressure and flow variables are obtained
using the developed Newton-Raphson based TPNS simulation model based on the
flow equations shown in Table 3.15, compressor equation (3.29) and the mass
balance equations shown in Table 3.16. The application of the Newton-Raphson
based TPNS simulation model for the branched network was conducted based on the
pipe data shown Table 3.17.
Table 3.17 Pipe data for single CS and five customers
Pipe nodes
Diameter [mm]
Length [km] Start node End node
0 1 900 100
2 3 900 80
3 D1 600 70
3 4 900 80
4 D2 600 90
4 5 900 70
5 D3 600 80
5 6 900 90
6 D4 600 70
6 D5 600 80
3.4.3 Case 3B: Looped Pipeline Network Module
The third type of transmission network system where the developed Newton-
Raphson based TPNS simulation model applied is looped system. In the case of
looped TPNS, gas pipelines diverge and re-converge when gas is transported from
the source to demand stations.
Figure 3.21 shows typical looped transmission network system with one
compressor station for transmitting gas to eight different customers. For this TPNS,
there are 16 pipes, 1 compressor station, 1 loop and 7 junctions. As a result, based on
Table 3.3, there will be 9 pressure variables and 17 flow variable to be determined.
82
There are 16 pipe flow equations, 1 compressor equation, 2 looping equations and 7
mass balance equations which form the 26 independent equations to analyze the
TPNS.
From the discussion in section 3.2.1, the pipe flow equations for single phase gas
flow are formulated and summarized as shown in Table 3.18.
Figure 3.21 Looped pipeline network system
83
Table 3.18 Summary of flow equations and functional representations
Pipe node Flow equation Functional representation Start node End node
0 1 2101
21
20 QKPP =− 0),( 111 =QPf
2 3 2223
23
22 QKPP =− 0),,( 2322 =QPPf
3 4 2334
24
23 QKPP =− 0),,( 3433 =QPPf
4 D1 2114
21
24 CDD QKPP =− 0),( 144 =CQPf
4 D2 2224
22
24 CDD QKPP =− 0),( 245 =CQPf
3 5 2445
25
23 QKPP =− 0),,( 4536 =QPPf
5 D3 2335
23
25 CDD QKPP =− 0),( 357 =CQPf
5 6 2556
26
25 QKPP =− 0),,( 5658 =QPPf
6 D4 2446
24
26 CDD QKPP =− 0),( 469 =CQPf
6 7 2676
27
26 QKPP =− 0),,( 67610 =QPPf
7 D5 2557
25
27 CDD QKPP =− 0),( 5711 =CQPf
7 8 2878
28
27 QKPP =− 0),,( 88712 =QPPf
8 D6 2668
26
28 CDD QKPP =− 0),( 6813 =CQPf
8 9 2989
29
28 QKPP =− 0),,( 99814 =QPPf
9 D7 2779
27
29 CDD QKPP =− 0),( 7915 =CQPf
9 D8 2889
28
29 CDD QKPP =− 0),( 8916 =CQPf
Based on the discussion on sections 3.2.2, the looping condition of the network
can be formulated for the given network as
( )30.3567
2667
556
2556
535
2435
523
2223
5
27
DQl
DQl
DQl
DQl
DQl
l
l +++=
The functional representation for equation (3.30) takes the form
( ) ( )31.30,,,, 7654217 =QQQQQf
The compressor equation for the TPNS is given as:
84
( ) ( ) ( ) ( )32.31]///[ 314
213121
2
1
2 ++++=⎟⎟⎠
⎞⎜⎜⎝
⎛nQAnQAnQAA
ZRTmn
PP
S
m
The corresponding functional representation for equation (3.32) takes the form
( ) ( )33.30,, 12118 =QPPf
The remaining mass balance equations for the network were formulated and
summarized as shown in Table 3.19.
Table 3.19 Summary of mass balance equations and functional representations
Node Mass balance equation Functional representation
2 721 QQQ += ( ) 0,, 72119 =QQQf
3 432 QQQ += ( ) 0,, 43220 =QQQf
4 213 CC QQQ += ( ) 0,, 21321 =CC QQQf
5 354 CQQQ += ( ) 0,, 35422 =CQQQf
6 465 CQQQ += ( ) 0,, 46523 =CQQQf
7 5876 CQQQQ +=+ ( ) 0,,, 587624 =CQQQQf
8 698 CQQQ += ( ) 0,, 69825 =CQQQf
9 879 CC QQQ += ( ) 0,, 87926 =CC QQQf
Based on the flow equations in Table 3.18, looping condition (3.31), the
compressor equation (3.33) and mass balance equations in Table 3.19, the
solutions to the unknown pressure and flow variables were obtained using the
developed Newton-Raphson based TPNS simulation model. The application of the
Newton-Raphson based TPNS simulation model for the looped network was
conducted based on the pipe data shown Table 3.20.
85
Table 3.20 Pipe data for single CS and eight customers
Pipe nodes
Diameter [mm]
Length [km] Start node End node
0 1 900 80
2 3 900 60
3 4 600 40
4 D1 600 20
4 D2 600 17
3 5 900 55
5 D3 600 18
5 6 900 70
6 D4 600 50
6 7 900 75
7 D5 600 40
7 8 900 184
8 D6 600 18
8 9 900 6
9 D7 200 33
9 D8 900 87
3.4.4 Case Study: Malaysia Gas Transmission System
In this section, the Newton-Raphson based TPNS simulation model was tested
with the actual data from the existing pipeline network system. The network was
identified since it consists of all the pipeline configurations, namely: gunbarrel,
branched, and looped pipeline network systems.
The objective of the case study was to have a comparison between the results of
TPNS simulation model and the actual data collected in the field. However, some
data, like the performance of the compressor and flow consumption by the customers
are considered confidential by the pipeline company. Hence, it would be difficult to
86
make actual comparison between the results of the simulation with that of the field
data. Several options were also tried in order to get the performance of the
compressor.
As an alternative to making an actual comparison between the results of the
TPNS simulation and the field data, the case study was applied to illustrate the
application of the TPNS simulation model to real pipeline network system. The detail
applications of the major subtasks of the TPNS simulation model were evaluated
based on the network system. Sensitivity analysis was also conducted on the basis of
the pipeline configuration system.
The TPNS considered for the case study consists of one compressor station with
two compressors working in parallel and having power capacity of 18 to 24 MW
each. The pipeline network system serves nine major power plant customers and one
Gas District Cooling (GDC) system. The two compressors are working in order to
satisfy the daily requirement of natural gas ranging from 0.8 to 1.7 billion standard
cubic feet per day (BSCFD) which is equivalent to the consumption of gas raging
from 22.65 to 48.14 million metric standard cubic meters per day (MMSCMD). Each
compressor has a capacity of 1 to 1.2 BSCFD which gives a cumulative capacity of
the compressors delivering 2 to 2.4 BSCFD to various customers which is equivalent
to 56.63 to 67.96 MMSCMD.
Due to the limitation of data regarding the performance map of the actual
compressors working in the field, gas pipeline compressors from [55] were used for
the analysis of the network. The compressor has a maximum capacity of 680m3/min
which is equivalent to 40800m3/hr (0.98 MMSCMD). The maximum speed is limited
to 10500rpm and the maximum head of the compressor is 108kJ/kg.
The gas flow rates in the pipe sections are 1Q , 2Q , … etc. and the flow rates
passing out the gas pipeline to various customers are 1CQ , 2CQ ,.. etc. Each customer is
identified by their customer station as 1D , 2D , etc. Figure 3.22 shows the network
87
configuration of part of the Malaysia gas pipeline network system considered for
case study.
There are 19 pipe elements connecting the various nodes with length of pipes
ranges from 6 km to 200 km. The diameter of the pipes ranges from 200 mm to 900
mm. Since the installation of the pipes was done on three different phases, the age of
the pipes is assumed to be ten years for determination of coefficient of friction of the
pipes. This TPNS consists of 1 loop and 8 junctions. As a result, based on Table 3.3,
there are 10 nodal pressures and 20 flow variables to be determined. Therefore, a
total of 30 independent equations were obtained in order to solve the network
problem. There are 19 pipe flow equations, 1 compressor equation, 2 looping
equations and 8 mass balance equations which formed the 30 independent equations
to analyze the TPNS. All the equations were formulated based on the discussion in
section 3.2, page 47. The results for unknown variables were determined on the
basis of multivariable Newton-Raphson algorithm.
88
Figure 3.22 Part of the existing pipeline network system
3.5 Enhanced Newton-Raphson based TPNS Simulation Model
The enhanced Newton-Raphson based TPNS simulation consists of additional
features such as two-phase flow analysis, internal corrosion and temperature
variations. Corrosion and two-phase flow analysis modified the flow equations.
When temperature variation was considered during analysis, it added (np+ ns)
temperature equations with equal number of unknowns. This section discusses two-
phase flow modeling, effect of internal corrosion and temperature variation in TPNS
simulation model.
3.5.1 Two-phase Gas-liquid Flow Modeling
As shown in equation (2.13), the total pressure gradient equation is composed of
three components: frictional, gravitational, and acceleration components. This
89
equation further developed here to enable the calculation of each of the pressure-
gradient components and the total pressure gradient. The basic facts and relationships
between the parameters are developed based on Figure 2.5.
The frictional pressure-gradient component is given by:
( )34.322 2'2'
NS
fmNS
F
Mf
Dvf
DdLdP
ρρ ==⎟
⎠⎞−
where fM is the mixture total mass flux, and mNSf vM ρ= for no-slip condition
and 'f is the fanning friction factor.
For rough pipes, ( )DReff m ε,'' = , and it can be determined from a Moody
chart [23] or from the following equation developed in [70].
( )35.3Re
101021001375.0316
4'
⎥⎥
⎦
⎤
⎢⎢
⎣
⎡
⎟⎟⎠
⎞⎜⎜⎝
⎛+×+=
Df ε
For smooth pipes, the friction factor is a function of the Reynolds number only,
( )Reff = , and a Blasius–type equation can be used as given [70]
( ) ( )36.3nReCf −=
where C is constant.
In equation (3.36), for laminar flow, the exponent n is 1 and that the values of
the constant are 16=C and 64=C for fanning and Darcy’s friction, respectively. For
turbulent flow, different values are available for different ranges of the Reynolds
number. For all practical purpose, the correlation covering the widest range of the
Reynolds number is 2.0=n with 046.0=C and 184.0=C for fanning and Darcy’s
friction respectively.
The gravitational pressure gradient component of the pressure drop for two-phase
flow can be determined as
90
( )37.3sinsin θρθρ ggdLdP
NSmG
==⎟⎠⎞−
The acceleration pressure gradient component is given as in [70]:
( )38.31)1(2
2
dLdA
AM
dPdx
dPdx
dLdP
dLdxM
dLdP P
PNS
fLGGLf
A ρυυυ −
⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡ −++=⎟
⎠⎞−
Therefore, the total pressure gradient can be obtained by combining the
frictional, gravitational, and acceleration pressure gradient component given in
equation (3.34), (3.37) and (3.38) respectively, which can be written as:
( )39.31)1(
sin2
22
2'
dLdA
AM
dPdx
dPdx
dLdP
dLdxM
gvfDdL
dP
P
PNS
fLGGLf
NSmNS
ρυυ
υ
θρρ
−⎭⎬⎫
⎩⎨⎧
⎥⎦⎤
⎢⎣⎡ −+++
+=−
Solving for the total pressure gradient from equation (3.39) yields
( )40.3)1(1
1sin2
2
222'
⎥⎦⎤
⎢⎣⎡ −++
−++=−
dPdx
dPdxM
dLdA
AM
dLdxMgvf
DdLdP
LGf
P
PNS
fGLfNSmNS
υυρ
υθρρ
Numerical evaluation
In order to see the effect of each component of the total pressure gradient,
numerical example in [70] is considered here.
The followings are detail of the oil and gas mixture considered for the analysis.
m.D 0510= , 010=θ , skgM L /5.1= , skgM G /015.0= . The temperature of the mixture is
assumed to be 250C and the physical properties of the fluids are: 3/5.1 mkgG =ρ ,3/850 mkgL =ρ , mskgL /100.2 3−×=μ , mskgG /100.2 5−×=μ .
91
Based on the above data and principles of homogeneous two-phase flow analysis,
the frictional pressure gradient can be evaluated based on equation (3.34) as:
mpadLdP
F/3.741=⎟
⎠⎞−
The gravitational pressure gradient can be obtained from equation (3.37) as:
mPadLdP
G/4.219=⎟
⎠⎞−
Similarly, the acceleration pressure gradient component can also be obtained
based on equation (3.38). Note that for constant x , the value of Lυ is also constant.
Furthermore, from the given information the cross-sectional area of the pipe PA is
constant. As a result, the acceleration pressure gradient equation (3.38) reduces to
( )41.32
dLdP
dPdxM
dLdP G
fA
υ=⎟
⎠⎞−
Based on equation (3.41) and assuming ideal gas law, the acceleration pressure
gradient is determined to be
mPadLdP
A/80.19=⎟
⎠⎞−
As a result the total pressure gradient is
mPadLdP /5.1980=⎟
⎠⎞−
Note that for the above homogeneous two-phase gas-liquid mixtures, the
acceleration pressure gradient component contributes only 2.02% to the total
pressure gradient. Hence, the acceleration pressure gradient is usually omitted from
92
calculation. Neglecting the effect of acceleration component of the pressure gradient,
equation (3.40) can be simplified for horizontal pipe flow as
( )42.32 2'mmvf
DdLdP ρ=−
As given in equation (3.36), the frictional coefficient for smooth turbulent flow
can be described as function of Reynolds number as
( )43.3046.0 2.0' −= mRef
The mixture Reynolds number for the two-phase flow is given
( )44.34D
QRem
mmm μπ
ρ=
The mixture velocity can also be expressed for two-phase flow as
( )45.342D
Qv mm π=
Substituting equations (3.43), (3.44), and (3.45) into equation (3.42) results
( )46.344046.02
2
2
2.0
⎟⎟⎠
⎞⎜⎜⎝
⎛
⎟⎟⎟
⎠
⎞
⎜⎜⎜
⎝
⎛
⎟⎟⎠
⎞⎜⎜⎝
⎛×=−
−
DQ
DQ
DdLdP m
m
mmm πμπ
ρρ
By integrating equation (3.46) from 0=L , 1PP = , to LL = , 2PP = we get
( )47.344046.02
2
2
2.0
21 LDQ
DQ
DPP m
m
mmm ⎟
⎟⎠
⎞⎜⎜⎝
⎛
⎟⎟⎟
⎠
⎞
⎜⎜⎜
⎝
⎛
⎟⎟⎠
⎞⎜⎜⎝
⎛×=−
−
πμπρ
ρ
For two-phase flow mixture, the pressure drop equation for any pipeline element
connecting node i and j shown in Figure 3.2, relating upstream pressure iP ,
downstream pressure jP , and the mixture flow through pipe mQ can be expressed as:
93
( )48.380.1mijji QKPP =−
where ijK is to be determined by the mixture properties and the
characteristics of pipe connecting node i and j .
ijK takes different forms depending on the dimensions of the parameters used
in the flow equations. When ][kPaP , ][kmL , ]/[ 3 hrmQm , ][mmD , ]/[ 3mkgmρ ,
]/[ smKgmμ , the expression for ijK takes the form
( )49.3100791.3 8.4
2.08
DLK
m
mmij
−
⎟⎟⎠
⎞⎜⎜⎝
⎛×=
μρ
ρ
If the pipeline networks system shown in Figure 3.3 is used to transmit two-
phase gas-liquid mixtures, the flow equations governing the simulation can be
generated for the pipes based on equation (3.48). Assuming that the source
pressure at node 0 and the demand pressures at node 6 and 9 are known, the two-
phase flow equations for the pipes and the corresponding functional representation
can be summarized as shown in Table 3.21. Note that for ease of simplification,
the mixture flow rate mQ and single phase gas flow rate Q are represented by
same variable Q .
Table 3.21 Two-phase flow equations and functional representations
Pipe node Two-phase flow equation Functional representation Start node End node
0 1 80.110110 QKPP =− 0),( 111 =QPf
2 3 80.112332 QKPP =− 0),,( 1322 =QPPf
3 4 80.123443 QKPP =− 0),,( 2433 =QPPf
3 7 80.133773 QKPP =− 0),,( 3734 =QPPf
5 6 80.125665 QKPP =− 0),( 255 =QPf
8 9 80.138998 QKPP =− 0),( 386 =QPf
94
The application of the enhanced Newton-Raphson TPNS simulation model for
two-phase flow analysis is demonstrated based on gunbarrel and branched network
configurations.
i) Case 1C: Gunbarrel Pipeline Network Module
When two-phase gas-liquid flows through the gunbarrel TPNS, the flow
equations were affected. It is assumed that only the gas flows to the compressor
station for compression due to severe effect of the liquid to the compressor stations.
In practice, scrubbers are usually installed in front of the compressor station to
remove any unwanted elements.
For the gunbarrel pipeline network system shown in Figure 3.19, the pipes flow
equations for two-phase gas-liquid flow are given based on equation (3.48) and
summarized as shown in Table 3.22.
Table 3.22 Summary of flow equations for the given gunbarrel TPNS
Pipe node Flow equation Functional representation Start node End node
0 1 80.10110 QKPP =− 0),( 11 =QPf
2 3 80.12332 QKPP =− 0),,( 322 =QPPf
4 5 80.14554 QKPP =− 0),,( 543 =QPPf
Similarly, based on the discussion on compressor stations modeling in section
3.2.3, the remaining compressor equations for the network can be formulated and
represented in functional forms as shown in Table 3.12.
Based on the flow equations shown in Table 3.22 and the compressor equations
shown in Table 3.12, the solution to the unknown pressure and flow variables were
obtained using the enhanced Newton-Raphson based TPNS simulation model.
95
ii) Case 2C: Branched Pipeline Network Module
The branched pipeline network discussed in 3.4.2 is considered here. Based on
equation (3.48), the pipe flow equations for two-phase gas-liquid flow were
formulated and summarized as shown in Table 3.23.
Table 3.23 Summary of flow equations and their functional representation
Pipe node Flow equation Functional representation Start node End node
0 1 80.110110 QKPP =− 0),( 111 =QPf
2 3 80.112332 QKPP =− 0),,( 1322 =QPPf
3 D1 80.11133 CDD QKPP =− 0),( 133 =CQPf
3 4 80.123443 QKPP =− 0),,( 2434 =QPPf
4 D2 80.12244 CDD QKPP =− 0),( 245 =CQPf
4 5 80.134554 QKPP =− 0),,( 3546 =QPPf
5 D3 80.13355 CDD QKPP =− 0),( 357 =CQPf
5 6 80.145665 QKPP =− 0),,( 4658 =QPPf
6 D4 80.14466 CDD QKPP =− 0),( 469 =CQPf
6 D5 80.15566 CDD QKPP =− 0),( 5610 =CQPf
The remaining equations from the compressor maps and mass balance were
formulated following the same procedure as in section 3.4.2. The compressor
equation for the TPNS is given as in equation (3.28) and represented in functional
form as in equation (3.29). The remaining mass balance equations for the network
were formulated and summarized as shown in Table 3.16.
The solutions to the unknown pressure and flow variables were obtained using
the developed enhanced Newton-Raphson based TPNS simulation model based on
the flow equations shown in Table 3.23, compressor equation (3.29) and the mass
balance equations shown in Table 3.16.
96
3.5.2 The Effect of Internal Corrosion
The effect of corrosion is to modify the flow equation during the development of
the governing simulation equations. Since corrosion in oil and gas industries is one
of the serious challenges which affect the performance of the pipeline network
system, the effect of corrosion on the performance of the gas transmission system
have to be analyzed and incorporated during the development of TPNS simulation
model.
So far, limited information is available from the literatures about how the
roughness of the pipes varies with the age of the pipe. In this section, a solution
scheme is developed to incorporate the age of the pipe to flow equation. Figure 3.23
shows the general procedure proposed to incorporate the effect of corrosion with age
of the pipe and flow modification.
Data of roughness of the coated pipe with service life of the pipe from [22] was
used for developing the correlation between the age of the pipe and roughness.
Several options were tried in order to get the best fit for the data of pipe roughness
against the service life of the pipe. It is observed that, even if the effect of pipe
roughness increase with increase in age of the pipe, the best fit for the data was
obtained when the approximation was done with 6 degree polynomial function with
correlation coefficient of 88.02 =R . However, this approximation gave negative
roughness values when the age of the pipe increased beyond the data points. A better
approximation for the given data was obtained when the data is approximated with
exponential function with correlation coefficient of 592.02 =R .
97
Figure 3.23 Proposed scheme for incorporating the age of the pipes in flow equations
Based on the data points, exponential function for approximating the relationship
between the roughness of the pipe with that of the age of the pipe is obtained to be:
( )50.300353.0 03802.0 yer =
where r is the roughness of the pipe in mm and y is the age of the pipe in year.
Most gas pipelines operate in the turbulent zone. For the fully turbulent zone,
American Gas Association (AGA) recommends using the following formula for
calculating the transmission factor F which is based on the relative roughness Dr
and independent of the Reynolds number [23] .
( )51.37.34 10 ⎟⎠⎞
⎜⎝⎛=
rDLogF
where F is the transmission factor, r is the roughness of the pipe and D is
the diameter of the pipe.
The relationships between the transmission factor F and friction factor f can be
express as [23]:
98
( )52.32f
F =
Substituting equation (3.52) in equation (3.51) and rearranging gives:
( )53.37.32
12
10 ⎥⎦
⎤⎢⎣
⎡⎟⎠⎞
⎜⎝⎛
=
rDLog
f
From equations (3.50) and (3.53), it can be seen that the friction factor is a
function of the age of the pipe. Figure 3.24 shows the effect of the years in service of
the pipe on friction factor for various diameters of the pipe.
Figure 3.24 The effect of length of service of the pipe on friction factor
The general flow equation for a single phase gas flow in equation (3.1) can be
modified based on the friction factor in equation (3.53) in order to incorporate the
effect of the age of the pipe on the pressure drop of the TPNS. For any pipe (Figure
3.2) connecting the upstream node i, downstream node j, and the flow through pipe
99
Q , the single phase flow equation with corrosion relating the two nodes can be
represented as
( )54.32'22 QfKPP ijji =−
where
( )55.3103599.42
58' L
TP
DGZTK
n
nij ⎟⎟
⎠
⎞⎜⎜⎝
⎛×=
If all the pressure and flow variables are unknown, equation (3.54) can also be
represented in functional form as
( )56.30),,( =QPPf ji
The application of the enhanced Newton-Raphson based TPNS simulation model
for incorporating the effect of internal corrosion is demonstrated based on gunbarrel
and branched network configurations.
i) Case 3C: Gunbarrel Pipeline Network Module
When corrosion is considered during the analysis of gas transmission system, the
overall effect is to modify the pressure drop equations. The gunbarrel transmission
network configuration shown in Figure 3.19 is considered here. For this pipeline
network, based on equation (3.54), the pipe flow equations for single phase gas flow
with corrosion were formulated and summarized as shown in Table 3.24.
Table 3.24 Summary of flow equations for the given gunbarrel TPNS
Pipe node Flow equation Functional representation Start node End node
0 1 2101
21
20 QfKPP =− 0),( 11 =QPf
2 3 2223
23
22 QfKPP =− 0),,( 322 =QPPf
4 5 2345
25
24 QfKPP =− 0),,( 543 =QPPf
100
Similarly, based on the discussion on compressor stations modeling in section
3.2.3, the remaining compressor equations for the network were formulated and
represented in functional forms as shown in Table 3.12.
Based on the flow equations shown in Table 3.24 and the compressor equations
shown in Table 3.12, the solution to the unknown pressure and flow variables were
obtained using the enhanced Newton-Raphson based TPNS simulation model.
ii) Case 4C: Branched Pipeline Network Module
For the branched pipeline network shown in Figure 3.20, based on equation
(3.54), the pipe flow equations for single phase gas flow with corrosion were
formulated and summarized as shown in Table 3.25.
Table 3.25 Summary of flow equations and their functional representation
considering corrosion
Pipe node Flow equation Functional representation Start node End node
0 1 21101
21
20 QfKPP =− 0),( 111 =QPf
2 3 21223
23
22 QfKPP =− 0),,( 1322 =QPPf
3 D1 21313
223 CDD QfKPP =− 0),( 133 =CQPf
3 4 22434
24
23 QfKPP =− 0),,( 2434 =QPPf
4 D2 22524
224 CDD QfKPP =− 0),( 245 =CQPf
4 5 23645
25
24 QfKPP =− 0),,( 3546 =QPPf
5 D3 23735
225 CDD QfKPP =− 0),( 357 =CQPf
5 6 24856
26
25 QfKPP =− 0),,( 4658 =QPPf
6 D4 24946
226 CDD QfKPP =− 0),( 469 =CQPf
6 D5 251056
226 CDD QfKPP =− 0),( 5610 =CQPf
The remaining equations from the compressor maps and mass balance were
formulated following the same procedure as in section 3.4.2. The compressor
101
equation for the TPNS is given as in equation (3.28) and represented in functional
form as in equation (3.29). The remaining mass balance equations for the network
were formulated and summarized as shown in Table 3.16.
The solutions to the unknown pressure and flow variables were obtained using
the developed enhanced Newton-Raphson based TPNS simulation model based on
the flow equations shown in Table 3.25, compressor equation (3.29) and the mass
balance equations shown in Table 3.16.
3.5.3 Modeling Temperature Variations
When the temperature of the gas varies due to heat transfer and compression, it is
essential to consider temperature variations during modeling. As presented in section
2.4.2, the variations of temperature within the TPNS are described based on
equations (2.14) and (2.15). These two equations should have to be integrated into
the governing simulation equations to represent the effect of temperature variations
for the enhanced Newton-Raphson based TPNS simulation.
Equation (2.14) is modified before it is integrated into the governing simulation
equations as the governing simulation equation takes only equations which are
functions of pressure, flow, and temperature variables. The mass flow rate M used
in equation (2.15) is given as function of volumetric flow Q and density of the gas as
( )57.3QM ρ=
Substituting equation (3.57) in equation (2.15) yields
( )58.3QCLUD
Pρπθ Δ
=
Therefore, equation (2.14) is modified for any pipe element (Figure 3.2)
connecting the upstream node i, downstream node j, and the flow through pipe Q , as
102
( )59.3)( /QTijsisj eTTTT −+=
where PC
LUDTijρ
π Δ=
If all the variables in equation (3.59) are unknown, it can also be represented in
terms of function form as
( )60.30),,( 21 =QTTf
The second equation to represent the variation of temperature in gas transmission
network system is equation (2.16). Since equation (2.16) is function of the required
variables, it was used without any modifications on it. If all the variables are
unknown in equation (2.16), the functional form of the equation takes the form
( )61.30),,,( 2121 =PPTTf
Equation (3.60) and (3.61) are the basic equations which govern the enhanced
Newton-Raphson based TPNS simulation under variable temperature conditions.
The application of the enhanced Newton-Raphson TPNS simulation model for
temperature variations is demonstrated based on the gunbarrel pipeline network
configurations shown in Figure 3.19. If the temperature of the gas is assumed to be
constant, there were 4 pressure variables and 1 flow variable to be determined for the
TPNS. When temperature variation is considered during the analysis, np+ns temperature equations with equal number of unknown temperature variables were
added to the governing simulation equations. As a result, there were five more
temperature equations and five unknown temperature variables introduced to the
system. The numbers of total unknown pressure, flow, and temperature variables
became ten.
Single phase flow model was used for the flow analysis. The summary of the
pipe flow equations for single phase gas flow model using general equation was
103
given as in Table 3.11. The remaining compressor equations for the network were
formulated and represented in functional forms as shown in Table 3.12.
The temperature equations were developed and represented in functional form
for the given pipeline network as shown in Table 3.26.
Table 3.26 Summary of temperature equations for gunbarrel TPNS
Temperature equation Functional representation
QTss eTTTT /01
01 )( −+= ( ) 0,16 =QTf
kk
PP
ZZ
TT
1
1
2
2
1
1
2
−
⎟⎟⎠
⎞⎜⎜⎝
⎛⎟⎟⎠
⎞⎜⎜⎝
⎛=⎟⎟
⎠
⎞⎜⎜⎝
⎛ ( ) 0,,, 21217 =PPTTf
QTss eTTTT /23
23 )( −+= ( ) 0,, 328 =QTTf
kk
PP
ZZ
TT
1
3
4
2
1
3
4
−
⎟⎟⎠
⎞⎜⎜⎝
⎛⎟⎟⎠
⎞⎜⎜⎝
⎛=⎟⎟
⎠
⎞⎜⎜⎝
⎛ ( ) 0,,, 43439 =PPTTf
QTss eTTTT /45
45 )( −+= ( ) 0,, 5410 =QTTf
Therefore, based on the flow equations in Table 3.11 , the compressor equations
in Table 3.12, and temperature equations in Table 3.26, the solutions to the unknown
pressure, flow, and temperature variables were obtained using the enhanced NR
based TPNS simulation model.
3.6 Performance Evaluation of the TPNS Simulation Model
As it is observed from the solution schemes discussed in the previous sections,
the successive substitution and Newton-Raphson methods provided solutions to the
unknown pressure, flow and temperature variables which are essential for evaluating
the performance of TPNS. The performance of the TPNS includes evaluation of the
flow capacities through each pipe elements, compression ratios at each compressor
stations, and the overall power consumption of the TPNS.
104
The flow capacities through each pipe are determined directly from the value of
the unknown variables obtained using either of the solution schemes discussed
above. The compression ratios (CR) at each compressor stations are evaluated based
on the pressure variables obtained by using the solution schemes. Based on the nodal
pressures, the compression ratio is defined as the ratio of discharge pressure to
suction pressure.
Based on the pressure and flow variables obtained, the simulation model is used
to evaluate the energy consumption of the TPNS. The amount of energy input to the
gas by the compressors is dependent upon the pressure of the gas and flow rate. The
power required by the compressor that takes into account the compressibility of gas
is given as [23]
( )62.31121
0639.4121
⎥⎥⎥
⎦
⎤
⎢⎢⎢
⎣
⎡−⎟⎟
⎠
⎞⎜⎜⎝
⎛⎟⎟⎠
⎞⎜⎜⎝
⎛⎟⎠⎞
⎜⎝⎛ +
⎟⎠⎞
⎜⎝⎛
−=
−kk
s
d
aS P
PZZQTk
kHPη
where HP is the compression power in kW.
3.7 Summary
The basic equations for the TPNS simulation are derived from the principles of
flow of fluid through pipe, compressor characteristics and the principles of mass
balance at the junction of the network. The existences of loops within the TPNS
create additional equations to the simulation model. These equations are developed
from the principle of looping conditions which states the pressure drop along any
closed loop is zero.
The solution for the unknown variables could be determined on the basis of an
iterative successive substitution or Newton-Raphson schemes. The method of
successive substitution is closely associated with the IFD where the governing
105
simulation equations are arranged in such a way that only one output is obtained
from the blocks. On the other hand, the Newton-Raphson solution scheme is
complex but powerful for the analysis of TPNS where the number of equations is
large.
The modeling of the application of iterative successive substitution scheme based
TPNS simulation model is demonstrated using two network configurations. The
development of the basic governing equations and the corresponding IFD generation
were discussed for the networks considered. The application of the Newton-Raphson
based TPNS simulation is also demonstrated using the three most commonly
networks configurations, namely: gunbarrel, branched and looped.
The next chapter discusses the results obtained by testing the developed TPNS
simulation model for different TPNS configurations. Table 3.27 shows the summary
of the different cases considered for the analysis.
106
Table 3.27 Various cases considered for the analysis using developed TPNS
simulation model
Model Case Module Section
Successive
substitution based
TPNS simulation
model
Case 1A 1CS 2 Customers 4.2.1
Case 2A 2CSs 2 Customers 4.2.2
Newton-Raphson
based TPNS
simulation model
Case 1B Gunbarrel 4.3.1
Case 2B Branched 4.3.2
Case 3B Looped 4.3.3
Malaysia TPNS
General 3.4.4
Enhanced Newton-
Raphson based TPNS
simulation model
Case 1C Gunbarrel 4.4.1
Case 2C Branched 4.4.1
Case 3C Gunbarrel 4.4.2
Case 4C Branched 4.4.2
The enhanced Newton-Raphson based TPNS simulation consists of additional
features such as two-phase flow analysis, internal corrosion and temperature
variations. Corrosion and two-phase flow analysis modified the flow equations.
When temperature variation is considered during analysis, it added temperature
equations with equal number of unknowns.
107
CHAPTER 4
RESULTS AND DISCUSSIONS
4.1 Introduction
In this chapter, the results of detail application of the TPNS simulation model for
various pipeline configurations are presented. The first part discusses the results of
the application of the TPNS simulation model based on successive substitution
solution scheme. The second part discusses the results of the application of the TPNS
simulation model based on the iterative Newton-Raphson solution scheme. The
application of the TPNS simulation model for real pipeline network system is also
discussed in this section. The third part discusses the results of enhanced Newton-
Raphson based TPNS simulation model. The fourth part of this Chapter discusses the
verification and validation of the simulation model.
4.2 Successive Substitution based TPNS Simulation Model
The results of the applications of successive substitution solution scheme for
TPNS simulation are demonstrated based on two different pipeline configurations
discussed in section 3.3.1 and 3.3.2. The successive substitution based simulation
model was implemented using excel based spreadsheet. The results of the analysis
were reported by Woldeyohannes and A. Majid [87].
108
4.2.1 Case 1A: Single Compressor Station and Two Customers Module
The single compressor station with two customers’ network shown in Figure 3.14
on page 66 is considered here for the analysis. Using the procedures described in
section 3.3.1, the results of the pressure and flow variables for this TPNS for the first
50 iterations are presented in Appendix C.
The convergences of the pressure and flow variables were studied. As stated in
[81], the calculation sequence dictated by the structure of the IFD determines
whether the sequence will converge or diverge. P3 and Q1 were the variables
identified from the IFD shown in Figure 3.15 on page 68 to start the analysis. When
the initial estimations for P3 and Q1 varied, there were two phenomena observed from
the simulation. The first case was the divergence case. In this case, the relative errors
calculated from each successive iterations followed irregular trends. The second case
was the slow convergence rate. In this case, when the initial estimations were far
from the final solutions of the variables, it took longer iterations to arrive the final
solution.
The convergence of the pressure variables for the TPNS is as shown in Figure
4.1. The pressure variables converged to their final solution with different rate of
convergence. The relative errors of the pressure variables started to become stable at
the 17th iteration for P2 and 16th iterations for P1 and P3.
109
Figure 4.1 Convergence of pressure variables (case 1A)
The convergences of the flow variables are shown in Figure 4.2. Around 20
iterations were required to get stable convergence graph for the flow variables.
However, QC1 and QC2 became stable starting from the 16th iterations as the initial
assumed values are near to these variables compared to Q1. The negative flow
observed for QC1 for the two iterations was due to the high initial estimation for P3.
110
Figure 4.2 The convergence of flow variables (Case 1A)
From the convergence graphs (Figure 4.1 and Figure 4.2), it is observed that, the
maximum relative percentage error after the 20th iterations is 1.21%. At the end of
the 50th iterations the maximum relative percentage error is 1.33E-04.
4.2.2 Case 2A: Two Compressor Stations and Two Customers Module
The two compressor stations two customers network shown in Figure 3.16 on
page 69 is considered here for the analysis. Using the procedures described in section
3.3.2, the results of the pressure and flow variables for this TPNS for the first 200
iterations.
The convergences of the pressure and flow variables were also studied. Figure
4.3 shows the convergence of the pressure variables for the first 200 iterations. The
corresponding convergences of the flow variables are shown in Figure 4.4. After the
20th iterations, the maximum relative percentage error observed was 32.7%. At the
end of the 50th iterations the maximum relative percentage error is 8.15E-02. These
111
show that as the number of unknown variables increased, the successive substitution
scheme took longer time to arrive at the specified tolerance as presented in [81].
Figure 4.3 The convergence of pressure variables (case 2A)
112
Figure 4.4 The convergence of flow variables (case 2A)
4.2.3 Limitations of Successive Substitution TPNS Simulation Model
Results from the numerical evaluation of the networks shown in Figure 3.14 on
page 66 and Figure 3.16 on page 69 revealed that the successive substitution based
TPNS simulation model could be applied to determine the flow and pressure
variables. However, it was observed that it took 50 iterations for case 1A and 200
iterations for case 2A to converge to relative percentage of errors of 1.33x10-4 and
8.15x10-4, respectively. As the number of pressure and flow variables increased, it
required more iteration to converge to the final solutions. Comparison of successive
substitution based TPNS simulation with that of Newton-Raphson based TPNS
simulation is reported by Majid and Woldeyohannes [88] based on two different
network configurations.
Although successive substitution scheme is straight forward technique and is
usually easy to program, sometimes the method may result in very slow convergence
rate depending on the IFD and the initial estimation for the unknown variables.
113
Hence, the method is limited only for TPNS with few numbers of pipes, compressor
stations and junctions. In addition, it is difficult to get the IFD of the given pipeline
network as the number of branches increases. Several correct IFD can be developed
for the given pipeline network system. However, it is very difficult to identify the
IFD which is convergent as discussed in [81]. For instance, for the TPNS shown in
Figure 4.5, one of the correct IFD is shown in Figure 4.6. It required testing many
alternatives to arrive at the correct IFD.
Figure 4.5 Pipeline network with three CSs serving four customers
The TPNS consists of 10 pipes, 3 compressor stations, 3 junction, and with no
loop. Therefore, np = 10, ns = 3, nl = 0, and nj = 3. As a result, there are 9 nodal
pressures and 7 flow parameters to be determined. Therefore, a total of 16
independent equations have to be obtained in order to solve the network problem.
The IFD is developed based on the governing simulation equations and shown in
Figure 4.6.
114
Figure 4.6 IFD for pipeline network with three CSs and four customers
4.3 Newton-Raphson based TPNS Simulation Model
The multivariable Newton-Raphson method which is discussed in section 3.4 on
page 72 is used for the analysis of TPNS with large number of pipes, compressor
stations, branches and loops. This section discusses results of detailed application of
the Newton-Raphson based TPNS simulation model using gunbarrel, branched and
looped configurations.
115
4.3.1 Case 1B: Gunbarrel Pipeline Network Module
The results and discussions presented in this section are based on the gunbarrel
TPNS network configurations shown in Figure 3.19 on page 77 and the methodology
presented in section 3.4.1 on page 76.
The network was analyzed with the aid of the developed Newton-Raphson based
TPNS simulation model. Gas pipeline compressors from [55] were used for the
analysis. The speed of the compressor was 8000 rpm. Note that the analysis could
also be done based on various speeds as long as the compressor speed is within the
operational limit. In order to make the multivariable Newton-Raphson convenient for
application, only single estimation that was used for all unknown variables was
required to start the iterations. The results for the unknown pressure variables and
flow variable were obtained with less than ten iterations. Table 4.1 shows the results
of the unknown nodal pressures and flow variable for the first ten iterations. Note
that pressure is measured in kPa and flow rate is measured in m3/hr.
Table 4.1 Results of nodal pressures and flow variables (case 1B)
Iteration P1 P2 P3 P4 Q Max. error 0 3000.00 3000.00 3000.00 3000.00 3000.00
1 2982.71 3591.98 3574.69 4183.96 2.40E+06 0.99875
2 1308.05 3665.21 2253.81 5210.26 1.49E+06 0.60892
3 2109.82 3691.36 2997.95 4638.76 987089 0.510994
4 2438.82 3545.31 3064.63 4384.63 715543 0.379496
5 2471.25 3505.5 3064.8 4346.88 638202 0.121187
6 2472.77 3505.01 3065.89 4345.74 632588 0.008874
7 2472.77 3505.01 3065.9 4345.73 632559 4.56E-05
8 2472.77 3505.01 3065.9 4345.73 632559 1.2E-09
9 2472.77 3505.01 3065.9 4345.73 632559 9.94E-16
10 2472.77 3505.01 3065.9 4345.73 632559 1.13E-16
116
The convergences of the pressure and flow variables were studied. Based on the
results obtained in Table 4.1, the convergence of nodal pressures and flow variables
for the network system are shown in Figure 4.7 and Figure 4.8, respectively.
Convergence was achieved in most of the initial estimations attempted for the
Newton-Raphson based TPNS simulation model. An initial estimation near to the
demand pressure requirements for both pressure and flow variables gave
convergences for the attempted simulation runs.
The convergence graphs shown in Figure 4.7 and Figure 4.8 indicated that the
relative percentage errors became stable starting from the 4th iteration. The high
value for Q that is observed at the second iteration in Figure 4.8 is due to the value of
initial estimation which is far off compared to the final solution.
117
Figure 4.7 Convergence of pressure variables (case 1B)
Figure 4.8 Convergence of flow variable (case 1B)
118
The results of Newton-Raphson based TPNS simulation model for the gunbarrel
network configurations are compared with the method proposed by Wu et al [47].
The pressure range for each node is between 689.5 kPa and 6895 kPa. There are 5
compressors within each compressor stations. The flow through pipes is assumed to
be 600 million standard cubic feet per day (MMSCFD) or equivalent to 707,921
million metric standard cubic meters per day (MMSCMD). The problem for the
network is the determination of nodal pressures which minimizes the fuel cost of the
compressors.
Wu et al [47] determined solutions for the optimal nodal pressures for
minimizing the fuel consumption of the system using exhaustive search method. The
results of the nodal pressures are shown in Table 4.2.
Table 4.2 Results of nodal pressures [47]
Node Pressure [kPa]
0 4112.88
1 3433.59
2 3640.43
3 2850.70
4 3088.85
5 2101.13
The pipeline network was also analyzed using the TPNS simulation model. In
TPNS simulation model, it is assumed that maintaining pressure requirement is
critical and therefore, the source pressure P0 and the demand pressure P5 are assumed
to be known. Therefore, the problem in TPNS simulation is finding the remaining
nodal pressures, flow rate, number of compressors within the stations, power
consumption and the speed of the compressors which satisfies the requirement.
A compressor map from the existing pipeline network system [89] was used for
the analysis. The characteristic map of the compressor was available in terms of the
119
discharge pressure versus flow capacity. As a result, the discharge pressure versus
flow capacity curves were approximated based on three degree polynomials.
For each compressor stations, identical compressors working with the same
speed are assumed. The simulation experiments were conducted by varying the
number of compressors and speed of the compressors in order to study the flow
capacity and the nodal pressures. From the simulation experiments, it was observed
that the minimum number of compressors required to meet the flow requirement was
8. Several speeds of compressor were tested for meeting the customer specifications.
The compressor speeds starting from 5550 rpm satisfied the requirement with various
power consumptions, nodal pressures and compression ratios.
It was observed that as the speed of the compressors increased, the results of
nodal pressures from the TPNS simulation were getting closer to the results of nodal
pressures obtained by Wu et al [47]. However, an increase in deviation of flow
parameter was observed when speed of compressors increased. After conducting
simulation experiments, compressor speed of 5775 rpm gave better results of nodal
pressures with maximum percentage error of 2.37% which was obtained at node 5 of
the network. The corresponding flow deviation was 10.71%. Figure 4.9 shows
comparison of the result of nodal pressures obtained from the TPNS simulation
model at speed of 5775 rpm and the results obtained in [47].
120
Figure 4.9 Comparison of nodal pressures
Wu et al [47] obtained the solution for nodal pressures so that the fuel
consumption for the system is minimized. The results obtained from TPNS
simulation model using the Newton-Raphson shows that the simulation model could
be used to compare various operations of the compressor. This could help to compare
the alternatives and select the one with minimum power consumption.
4.3.2 Case 2B: Branched Pipeline Network Module
The results and discussions presented in this section are based on the branched
TPNS network configurations shown in Figure 3.20 on page 79 and the methodology
presented in section 3.4.2 on page 79. The analysis was done based on source
pressure of 3500kPa and demand pressure requirement 4500kPa. The results and
analyses on the network were reported by Woldeyohannes and Majid [90].
121
The network was analyzed using TPNS simulation model based on Newton-
Raphson solution method. Two compressors were working for the system. The
characteristics map of the compressors were taken from Kurz and Ohanian [59]. For
the specified condition, it was assumed that the speed of the compressor is 8800 rpm.
Note that the analysis could also be done at any arbitrary speed within the working
limits of compressors. The results for the unknown pressure variables and flow
variables were obtained with less than ten iterations with relative percentage error of
1.76979E-15. Figure 4.10 shows the convergence of some of the nodal pressures to
the final pressure solutions for the first ten iterations. The convergences of the
remaining nodal pressure variables followed the same trends. The convergence of the
corresponding flow parameters are shown in Figure 4.11 and Figure 4.12.
122
Figure 4.10 Convergence of pressure (case 2B)
Figure 4.11 Convergence of main flow variables (case 2B)
123
Figure 4.12 Convergence of lateral flow variables (case 2B)
4.3.3 Case 3B: Looped Pipeline Network Module
The results and discussions presented in this section are based on the looped
TPNS network configurations shown in Figure 3.21 on page 82 and the
methodology presented in section 3.4.3 on page 81. The network was analyzed based
on single phase gas flow analysis. The pressure at node 0 was 3500kPa and the
demand pressure from the customer was assumed to be 4000kPa. The results of the
analysis using TPNS simulation for looped network configuration were reported by
Woldeyohannes and Majid [91].
124
The rate of convergence to the final solution depends on the initial estimation
and usually high at the beginning of the iterations. Based on the pipe data shown in
Table 3.20 and gas compressor data from [55], the results of the unknown pressure
and flow variables were obtained for the first ten iterations. The analysis was
performed based on compressor speed of 8500 rpm. For the initial estimations of
4000kPa for unknown pressure variables and 4000m3/hr for unknown flow
variables, the results from the TPNS simulation model using Newton-Raphson
method at the end of the 10th iterations are shown in Table 4.3 and Table 4.4. The
corresponding compression ratio and power consumption for the system were
obtained to be 1.49837 and 14.1147 MW, respectively. At the end of the 10th
iteration, the maximum percentage relative error was 7.197x10-11.
Table 4.3 Results of nodal pressures after ten iterations
Node Pressure [kPa]
0 3500.00
1 2880.47
2 4316.02
3 4136.33
4 4014.28
5 4064.4
6 4047.89
7 4046.37
8 4006.1
9 4005.68
125
Table 4.4 Results of main and branch flow variables after 10 iterations
Main flow [m3/hr] Branch flow [m3/hr]
Q1 740381 QC1 88740.9
Q2 529934 QC2 96253
Q3 184994 QC3 199260
Q4 344940 QC4 102997
Q5 145679 QC5 113299
Q6 42682.8 QC6 61105.3
Q7 210447 QC7 2564.04
Q8 139831 QC8 76161.8
Q9 78725.9 - -
The convergences of the pressure variables at node 1 and 2 and the main flow
variable 1Q were studied under various initial estimations. Figure 4.13 and Figure
3.14 show the convergence of nodal pressures to the final pressure solutions for the
first ten iterations at node 1 and 2 for initial estimations of 3000, 4000 and 10000kPa.
The convergences of the remaining nodal pressures followed the same trend as that
of the pressure at node 2.
126
Figure 4.13 Convergence of P1 for the first ten iterations for the initial estimations
(case 3B)
Figure 4.14 Convergence of P2 for the first ten iterations for the initial estimation
(case 3B)
127
The convergence of the main flow variable 1Q for the initial estimations of 3000,
4000 and 10000 m3/hr is shown in Figure 4.15. The simulation model produced
solutions for the unknown variables with wide range of initial estimation.
Figure 4.15 Convergence of Q1 for the first ten iterations for the initial estimations
(case 3B)
The results of the TPNS simulation model for single phase gas flow analysis
were compared with the method proposed by Osiadacz [7] based on the looped
network configuration shown in Figure 4.16. The pipeline details and the demand
requirements for the TPNS are shown in Table 4.5 and Table 4.6, respectively. For
this problem, Osiadacz applied Newton loop-node method in order to get the flow
and pressure variables by assuming fixed pressure ratios for each compressor.
However, the final nodal pressures obtained after achieving the predefined error
limits failed to satisfy the previously assumed pressure ratios. For instance, the
128
pressure ratio at CS1 was assumed to be 1.8 and that of CS2 was assumed to be 1.4.
However, the pressure ratios after the solutions were obtained were actually 1.34 and
1.1832 for CS1 and CS2, respectively.
Figure 4.16 Pipeline networks with 10 pipes and two compressor stations [7]
Table 4.5 Pipe data [7]
Pipe nodes
Diameter [mm]
Length [km] Start node End node
1 2 700.00 70.00
1 3 700.00 60.00
2 3 700.00 90.00
2 4 600.00 50.00
3 6 600.00 45.00
5 8 600.00 70.00
7 9 600.00 80.00
8 9 500.00 70.00
8 10 500.00 45.00
9 10 500.00 75.00
129
Table 4.6 Pressure and flow requirements [7]
Node Flow [m3/hr] Pressure [kPa]
1 - 5000
2 20000 -
3 20000 -
4 0 -
5 0 -
6 0 -
7 0 -
8 15000 -
9 30000 -
10 45000 -
The simulation experiments were conducted by varying the number of
compressors and speed of compressors. The analysis was conducted using the
performance characteristics of the compressors taken from [59]. Similar compressors
are assumed to work on both compressor stations. From the simulation experiments,
it was observed that one compressor for each compressor station is sufficient to meet
the customer requirements.
Simulation experiments based on the given compressors characteristics showed
that, the two compressor stations have to work nearly with the same compression
ratios. An increase in compression ratio for the first station could result more flow
through pipe 1 and reversal flow through pipe 3. This might cause flow reversal in
the second compressor station. For instance, the compressor ratio mentioned for CS2
by Osiadacz [7] was achieved when the compressor runs with speed of 4800rpm.
Based on this reference speed for CS1, the speed of the compressor increased to
improve the compression ratio of CS2. The compressor speed at CS1 was increased
till flow reversal starts to CS (i.e. 5450 rpm). It was observed that, the deviations of
the nodal pressures and flow variables increased as the speed increase towards the
speed of 5450 rpm.
130
After conducting simulation analyses based on the requirements, compressors
speeds of 5025 rpm for CS1 and 4750 rpm for CS2 gave results of nodal pressures
and flow variables close to Osiadacz [7]. Mean absolute percent error of 5.10% was
observed between the two methods. The variations of the flow and nodal pressure
variables could be as a result of the type of flow equations that were used in the
analysis and the oversimplification of compressor stations in the case of the method
in [7]. Panhandle ‘A’ flow equation was used in [7] where as general flow equation
has been applied in the developed TPNS simulation. As developed in section 3.2.3,
the TPNS simulation model consists of detailed characteristics of the compressor
rather than only limited to compression ratio.
The results of comparison of the TPNS simulation model and that of Osiadacz
[7] for flow rate through pipes is shown in Figure 4.17. From the figure, it is
observed that the maximum deviation between the results of TPNS simulation model
and the results from Osiadacz [7] occurred at pipe 1, 4 and 6. This is as a result of the
higher compression ratio assumed at CS1 which was far from the calculated value.
The negative flow observed in pipe 3 of the network showed that the actual flow
direction for the gas is opposite to the assumed direction. Hence, the actual flow in
pipe 3 is from node 3 to node 2.
131
Figure 4.17 Comparison of flow rates between TPNS simulation and results of
Osiadacz [7]
The comparison of the corresponding nodal pressures between the two methods
is shown in Figure 4.18. From the figure, it is observed that higher deviation at nodal
pressures between the Osiadacz [7] and the results from the developed TPNS
simulation model happened at node 4. This is as a result of the value of the CR
assumed at CS1. Generally, the results of nodal pressure from the TPNS simulation
model is higher than the nodal pressures obtained from Osiadacz [7]. This could be
as a result of the flow equations used in both methods. The general flow equation
which was used in TPNS simulation model result less pressure drop compared to the
Panhandle ‘A’ flow equation which was used by Osiadacz [7].
132
Figure 4.18 Comparison of nodal pressures between TPNS simulation and results of
Osiadacz [7]
4.3.4 The Malaysia Gas Transmission Network Case Study
In this section, the TPNS simulation model using the Newton-Raphson method
was tested with the actual data from the existing pipeline network system. The
network was identified since it consists of all the pipeline configurations, namely:
gunbarrel, branched, and looped pipeline network systems.
Based on the data from the existing pipeline network system presented in section
3.4.4 on page 85, the TPNS simulation was evaluated. The various subtasks of the
simulation model were evaluated on the basis of the data taken from the field. The
main modules of the TPNS simulation model include input parameter analysis,
function evaluation module, and network evaluation module. Single phase flow
analysis at constant gas flow temperature is assumed for the evaluation of the TPNS
simulation model with the existing pipeline network system.
133
i) Input Parameter Analysis
This phase of the TPNS simulation consists of taking the input from the user and
making analysis in order to make the data suitable for the next phase of the
simulation. The input to the simulation includes pipe data, compressors data and
customer requirements.
a) Pipe Input Data
The TPNS under consideration for the case study consists of 19 pipes with
various diameter and length. The first step of the simulation is to take the relevant
input regarding the number of pipes involved, diameter of the pipes, length of the
pipes and age of the pipes. Using the input information, the TPNS simulation
analyzes the pipe data to determine the friction factor of each pipes as well as the
pipe flow resistance ijK which was defined in equation (3.1) on page 47. Based on
the data of the pipeline network system under consideration, the flow resistance and
friction factor for each pipe were determined and shown in Table 4.7.
134
Table 4.7 Results of input analysis for pipes
Pipe nodes Friction factor f
Flow resistance
ijKStart node End node
0 1 0.007407 7.62395E-06
2 3 0.007407 5.71796E-06
3 D1 0.007878 4.61777E-05
3 D2 0.007878 4.61777E-05
2 4 0.007407 5.71796E-06
4 5 0.007878 3.07851E-05
5 D3 0.007878 1.53926E-05
5 D4 0.007878 1.30837E-05
4 6 0.007407 5.24147E-06
6 D5 0.007878 1.38533E-05
6 7 0.007407 6.67096E-06
7 D6 0.007878 3.84814E-05
7 8 0.007407 7.14746E-06
8 D7 0.007878 3.07851E-05
8 9 0.007407 1.75351E-05
9 D8 0.007878 1.38533E-05
9 10 0.007407 5.71796E-07
10 D9 0.009403 0.00736663
10 11 0.007407 8.29105E-06
b) Compressor Input Data
After the preprocessing of the pipeline information has been completed, the next
step of the TPNS simulation is to take the data related to compressor station as an
input. The data related to compressors consists of the performance map of the
compressors, number of compressors working within the station and the compressor
speed. Based on the performance map of the compressor used, the Newton-Raphson
based TPNS simulation model carried out the analysis to determine the coefficient of
135
the mathematical equation which represents the characteristics of the compressor.
The coefficients were determined as discussed in section 3.2.3 on page 52.
c) Customer Requirement
One of the basic inputs for the TPNS simulation model is to take data related to
the pressure requirement of the customers. The customer specification in terms of
pressure requirement is usually given to satisfy the remote customer pressure
requirement. This is because, if the pressure at the end point is satisfied, the entire
system will have sufficient pressure requirement. For the case study under
consideration, the maximum pressure at the end point of the TPNS could reach
6800kPa and the minimum pressure is limited 4000kPa. The pressures at the end of
the pipe joining the customer with the system will be regulated to the customer
requirement. The source pressure varies from 3000kPa to 3500kPa.
d) Initial Estimations and Maximum Error Limit
The TPNS simulation requires an initial estimation of the unknown variables.
The relative error introduced depends on the initial estimation of the unknown
variables. The simulation model was tested by giving wide range of initial
estimations for the unknown variables and convergences were achieved in most of
the trails. Proper estimation for the nodal pressures could be obtained from the exit
pressure requirements at the various demand stations. Furthermore, proper estimation
for the flow variables could be obtained from the performance characteristics of the
compressor provided for the simulation.
ii) Output of TPNS Simulation
The final results from the TPNS simulation for the required nodal pressure and
flow through pipes depend on the number of iterations or the predefined percentage
error limit. After the predefined number of iteration or the maximum relative
percentage error for the unknown variables was satisfied, the TPNS simulation gave
136
the value of each of the unknown variables, compression ratio, and power
consumption for the system. The results were displayed on the basis of iterations.
The rate of convergence to the final solution depends on the initial estimation and
usually high at the beginning of the iterations. Based on the pipe data shown in
Table 4.7, source pressure of 3500kPa and an end pressure requirement of 4000kPa,
compressor speed of 7600 rpm, the results of the unknown nodal pressure and flow
variables were obtained for the first ten iterations. At the end of the 10th iteration, the
maximum relative percentage error was obtained to be 8.66347E-17. The results
from the TPNS simulation for the unknown nodal pressures and flow variables are
shown in Table 4.8. The corresponding compression ratio and power consumption
for the system were obtained to be 1.4951 and 10.9378 MW, respectively.
Table 4.8 Results of nodal pressures and flow variables after 10 iterations
Node Pressure[kPa] Main flow m3/hr Branch flow m3/hr
0 3500.00 Q1 770480.0 QC1 135404.0
1 2779.23 Q2 142038.0 QC2 135404.0
2 4155.23 Q3 357634.0 QC3 59865.3
3 4104.47 Q4 270808.0 QC4 64933.0
4 4066.28 Q5 124798.0 QC5 134465.0
5 4006.89 Q6 232836.0 QC6 69509.2
6 4031.19 Q7 98371.1 QC7 76459.2
7 4023.17 Q8 28861.9 QC8 41252.3
8 4022.43 Q9 94440.5 QC9 1726.45
9 4002.95 Q10 53188.2 - -
10 4002.74 Q11 51461.8 - -
The convergences of the pressure variables at node 1 and 2 and the main flow
variable 1Q were studied under various initial estimations.
Figure 4.19 and Figure 4.20 show the convergence of nodal pressures to the final
solutions for the first ten iterations at node 1 and 2 of the pipeline network system
137
shown in Figure 3.22 based on the initial estimations of 2000, 4000 and 10000kPa.
The convergence of the remaining nodal pressures can also be plotted following the
same procedures as that of the nodal pressures at node 1 and 2.
Figure 4.19 Convergence of P1 for the first ten iterations for the initial estimations
138
Figure 4.20 Convergence of P2 for the first ten iterations for the initial estimations
From the study of the convergence of the nodal pressures at node 1 and 2, it was
observed that the TPNS simulation model could provide solutions to pressure
variables with a wide range of initial estimations. From the simulation experiments,
the initial estimation near to the end pressure requirement was obtained to be a good
initial estimation and resulted solutions in all the tests conducted. The convergence
of the main flow parameter 1Q for the initial estimations of 2000, 4000 and
10000m3/hr is shown in Figure 4.21.
139
Figure 4.21 Convergence of Q1 for the first ten iterations for the initial estimations
The convergence of the flow variables through the main pipes and the lateral
pipes were also plotted. Figure 4.22 shows the convergence of the remaining main
flow variables for the first ten iterations. As indicated in the figure, almost all the
main flow variables solutions were obtained starting from the third iteration.
140
Figure 4.22 Convergence of the main flow variables
iii) Application of TPNS Simulation for Compressor Performance Analysis
Further simulation study was conducted in order to apply the TPNS simulation
model for performance analysis of the compressor used for the pipeline network
system shown in Figure 3.22. For the performance analysis of the compressor,
suction pressure (P1), discharge pressure (P2) and the flow rate through the
compressor (Q1) were considered. The performance of the compressor was studied
for source pressure of 3500kPa to meet pressure requirements of various demand
pressures ranging from 4000 to 5000kPa. The analysis was performed with
compressor speeds of 5000, 5500, 6000 and 6500rpm. Note that the analysis can also
be made at any speed of the compressor as long as the speed is within the working
limit of the compressor. The results of the application of the Newton-Raphson based
TPNS simulation for performance analysis was reported in [92].
141
The variation of the discharge pressure (P2) of the compressor with flow rate (Q1)
is shown in Figure 4.23. As it is observed from the figure, for a constant speed
operation of the compressor, an increased in discharge pressure gave rise to a
decrease in flow capacity of the system and vice versa. This showed that the
characteristic map generated using the TPNS simulation model is similar to the
characteristics maps of the compressors described in [60, 93]. As a result, the
simulation model could be used to analyze the performance of the compressors.
Figure 4.23 Discharge pressure variations with flow for various speeds
The variation of CR with flow and speed is shown in Figure 4.24. As it is seen
from the figure, higher CR for the system was achieved at lower flow capacities.
This is because, based on equation (3.76), power consumption of the system is a
function of CR, flow rate and other properties of the gas. From this equation, it is
observed that lower flow rate results higher CR values. For a constant flow
operation, an increase in speed of the compressor increased the CR of the system.
142
The results in the figure also showed the characteristic map of the compressor based
on CR which is similar in trend with that of the characteristic map plotted in [60].
Figure 4.24 Compression ratio variations with flow for various speeds
Power consumption variation with flow rate through the compressor based on
various speed is shown in Figure 4.25. It is observed that an increase in flow rate
increased the power consumption. For a constant flow operation, an increase in speed
of the compressor increased the power consumption of the system. As it is given in
[23], power consumption by the system is function of the flow rate, CR and
properties of the gas. When the speed of the compressor increased at constant flow
operation, it is shown in Figure 4.24 that the CR of the compressor is also increased.
Hence, the power consumption of the system increased due to an increment in CR of
the system.
143
Figure 4.25 Variation of power consumption with flow for various speed
4.3.5 The Case of Divergent in TPNS Simulation Model
The Newton-Raphson based TPNS simulation model plays significant role in
achieving the required pressure, flow, and temperature variables usually within four
to six iterations depending on the error tolerance limit. However, the Newton-
Raphson solution scheme has limitations depending on the initial estimations as
reported in [81]. Initial estimations which are too far from the final solution usually
result in divergence. In order to guide for successful convergence, several attempts
were conducted on TPNS simulation model and an initial estimations which gives
solution for the variables were identified. Initial estimations near to the demand
pressure requirements resulted in convergence for the attempts made on TPNS
simulation model. Sample of the divergent case for branched TPNS is shown in
Appendix D.
144
4.4 Enhanced Newton-Raphson based TPNS Simulation Model
The results and discussions in this section are based on the enhanced Newton-
Raphson based TPNS simulation model presented in section 3.5 on page 88. The
model was applied for analyzing gunbarrel and branched pipeline network
configurations considering two-phase gas-liquid flow analysis, single phase with
corrosion effect and single phase with temperature variations.
4.4.1 Two-phase Gas-liquid Flow Analysis
The analyses conducted in this section for the gunbarrel and branched network
configurations were based on the methodology discussed in section 3.5.1 on page 88.
For the analysis of the networks, it was assumed that the liquid holdup is
005.0== LLH λ which is very common in transmission network system [61]. As a
result, the density of the mixture and the viscosity of the mixture were determined to
be 3/7425.5 mkgm =ρ and mskgEm /0599.2 −=μ , respectively.
i) Case 1C: Gunbarrel Pipeline Network Module
The results and discussions presented in this section are based on the gunbarrel
TPNS network configurations shown in Figure 3.19 on page 77. The analysis is
based on the pipe data shown in Table 3.14 and the nodal pressure requirements data
shown in Table 3.13. The network was analyzed based on the enhanced Newton-
Raphson based TPNS simulation model considering effect of two-phase flow
analysis. The same compressors as in section 4.3.1 with speed of 8000 rpm were
used for the analysis. The results for the unknown pressure variables and flow
variable were obtained with less than ten iterations. The convergences of nodal
pressures and flow variable for the network system were studied. Based on the
results obtained from the iterations, the convergence of the pressure variables and
flow variable are as shown in Figure 4.26 and Figure 4.27, respectively.
145
Figure 4.26 Convergence of pressure variables (Case 1C)
Figure 4.27 Convergence of flow variable (Case 1C)
146
ii) Case 2C: Branched Pipeline Network Module
The results and discussions presented in this section are based on the branched
TPNS network configurations shown in Figure 3.20 on page 79. The network was
analyzed using the enhanced Newton-Raphson based TPNS simulation model
considering the effect of two-phase flow. The analyses regarding the network were
based on the pipe data shown in Table 3.17. The same compressor specifications
with speed 7800rpm was used as in 4.3.2. The pressure at node 0 was 3500kPa and
the demand pressure from the customer was assumed to be 4000kPa.
The results for the unknown pressure variables and flow variables were obtained
with less than ten iterations with relative percentage error of 8.16639E-17. Figure
4.28 shows the convergence of nodal pressures to the final pressure solution for the
first ten iterations with an initial estimation of 2000kPa for all pressure variables and
2000m3/hr for all flow variables. The simulation model was also tested by giving
wide range of initial estimations and convergence was achieved in most of the trials.
An initial estimations near to the demand pressure resulted in convergences in all the
tests conducted on the pipeline network system. Most of the pressure parameters
converged to their final solution after the third iterations.
147
Figure 4.28 Convergence of pressure variables (Case 2C)
The flow through main pipe joining node 5 and node 6 is less compared to the
other main pipe flows. As a result, the pressure drop from node 5 to node 6 is small
compared to the other main pipes. Hence, the convergence of the nodal pressure at
node 5 and node 6 coincides nearly to the same line. The result from the simulation
model showed that the pressure at end of the tenth iteration at node 5 and 6 were
4013.48 kPa and 4008.43kPa, respectively. The convergence of the corresponding
flow variables are shown in Figure 4.29 and Figure 4.30.
148
Figure 4.29 Convergence of main flow variables (Case 2C)
Figure 4.30 Convergence of lateral flow variables (Case 2C)
149
The effect of the variation of the liquid holdup on nodal pressures and flow
through pipes were also studied. Figure 4.31 shows the effect of liquid holdup on
nodal pressures. Since the customer requirements have to be satisfied in terms of
pressure, the nodal pressures remain nearly the same for various liquid holdups.
However, slight increment in pressure drop was observed when the liquid holdup
increased.
Figure 4.31 Effect of liquid holdup on nodal pressures
Liquid holdup had a significant effect on flow through pipes. An increase in
holdup decreased the flow through pipes as shown in Figure 4.32. This is because,
when the liquid holdup increased, flow resistance increased to result in a decrease in
flow of the mixture through pipes.
150
Figure 4.32 Effect of liquid holdup on flow through main pipes
4.4.2 The Effect of Internal Corrosion
The analyses conducted in this section for the gunbarrel and branched network
configurations were based on the methodology discussed in section 3.5.2 on page 96.
i) Case 3C: Gunbarrel Pipeline Network Module
The gunbarrel pipeline network shown in Figure 3.19 on page 77 was also
analyzed with the aid of the enhanced Newton-Raphson based TPNS simulation
model by considering the effect of corrosion. The same specifications of compressors
were used as in section 4.3.1. The age of the pipes could vary to give the
corresponding friction coefficient which results different pressure drop. The results
of the unknown nodal pressures and flow variable were obtained for the first ten
iterations for ten years old pipes. Based on these results, the convergence of nodal
151
pressures and flow variables for the network system are shown in Figure 4.33 and
Figure 4.34, respectively.
The comparison of the convergence graphs obtained by neglecting the effect of
corrosion in Figure 4.7 is identical to the convergence graph obtained considering the
effect of corrosion shown in Figure 4.33. However, for the case of corrosion large
shootings are observed on each graph from the initial estimation. This is resulted
from the behavior of the convergence in Newton-Raphson solution technique where
the graphs always tend towards the solution depending on the initial estimations.
Initial estimations less than the final solutions are always resulted the convergence
graphs to shoot upward before going stable at the solution. On the other hand, an
initial estimation which are greater than the final solution resulted the convergence
graphs to shoot downward before coming to stable at the final solutions.
Figure 4.33 Convergence of pressure variables with corrosion (Case 3C)
152
Figure 4.34 Convergence of flow variable with corrosion (Case 3C)
The results shown in Table 4.1 on page 115 were compared with the results
obtained by considering the effect of corrosion for various ages of the pipe. The
pressures at node 0 and 5 were constant throughout the various ages. This is because
the pressures at these nodes are the given conditions. The speed of the compressor is
assumed to be 8000 rpm for both cases. For the case of neglecting the effect of
corrosion, only the natural frictional coefficient of the pipe was considered which is
constant throughout the service of the pipes. However, for the case of pipes where
carrions is taken into account, the friction factor varied with service life of the pipe
based on the relationships developed in section 3.5.2 on page 96. The frictional
coefficient had an effect on the pressure drop, flow capacity, compression ratio and
power consumption.
Table 4.9 shows the comparison of the pressure variables at different nodes for
various ages of the pipes. As indicated in the table, higher pressure drops were
observed as the ages of the pipe increased. For instance, for pipe joining node 0 and
153
1, the pressure drop for the 20 years old pipe was higher than the pressure drop for
10 years old.
Table 4.9 Comparison of nodal pressures for different pipe ages
Nodes
Pressure [kPa] for various ages
New pipe 10 years
old pipe
20 years
old pipe
0 3000.00 3000.00 3000.00
1 2472.77 2460.22 2447.55
2 3505.01 3504.79 3504.51
3 3065.90 3055.53 3045.02
4 4345.73 4352.85 4359.99
5 4000.00 4000.00 4000.00
The flow capacity of the system was also studied for various ages of the pipes.
The variation of the flow capacity of the pipe with age of the pipe is shown in Figure
4.35. From the figure, it is observed that the capacity of the pipe is reduced as the
ages of the pipe increased. This is because, as the age of the pipes increase the
roughness of the pipe increases due to corrosion and accumulation of various
elements. As a result, the friction factors for the pipes increased which increase the
pipe flow resistance. An increase in flow resistance of the pipe decreases the flow
capacity of the system.
154
Figure 4.35 Variation of flow with age of the pipe
The variation of CR with ages of the pipe is shown in Figure 4.36. CR increased
as the age of the pipe increased. This is due to the fact that, an increase in ages of the
pipe could result in an increase in roughness of the pipe which leads to increase in
friction factor. As a result, the pressure drop increased which cause higher CR.
155
Figure 4.36 Variation of compression ratio with age of the pipe
The variation of the power consumption by the system with age of the pipe is
shown in Figure 4.37. As it is observed from the figure, for constant speed operation,
power consumption decreases with an increase in age of the pipes. As shown in
equation (3.62), power consumption of the system is mainly function of the flow
rate, CR and properties of the gas. From Figure 4.35, it is observed that an increase
in age of the pipe caused a decrease in flow capacity of the system. Moreover, it is
observed in Figure 4.36, CR increased with age of the pipes. Therefore, the trend of
power consumption with age of the pipes depends on either flow rate or CR. From
Figure 4.37, it is observed that the trend of power consumption followed the same
trend as that of Q1. This showed that the effect of flow variation was more dominant
than the effect of CR variations with age of the pipes on power consumption.
156
Figure 4.37 Variation of the power consumption with age of the pipes
ii) Case 4C: Branched Pipeline Network Module
The branched pipeline network shown in Figure 3.20 on page 79 was also
analyzed with the aid of the enhanced Newton-Raphson based TPNS simulation
model by considering the effect of corrosion. The same specifications of compressors
were used as in section 4.3.2. The age of the pipes could vary to give the
corresponding friction coefficient. The convergences of the pressure and flow
variables were also studied based on ten years old pipes. The convergence of some of
nodal pressure variables is shown in Figure 4.38. It is observed from the graph that
the solutions for the nodal pressures were obtained after the third iterations. The
maximum error at the end of the 10th iterations was 1.92x10-15. Figure 4.39 shows the
convergence of flow variables through main pipes. As it is observed from the graph,
the TPNS simulation nearly arrived to the final solution at the 4th iteration. At the end
of the 4th iteraion, the maximum relative error was 0.045935.
157
Figure 4.38 Convergence of pressure variables (Case 4C)
Figure 4.39 Convergence of flow through main pipes (Case 4C)
158
The convergence of flow variables through the branch pipes is shown in Figure
4.40. Similar to the convergence of the nodal pressures and flow variables through
main pipes, the TPNS simulation converged almost to the final solution at the end of
the 4th iteration for the flow variables through lateral pipes.
Figure 4.40 Convergence of flow through lateral pipes (Case 4C)
The results obtained in section 4.3.2 were compared with the results obtained by
considering the effect of corrosion for various ages of the pipe. Three groups of
pipes i.e. pipes with ages of 0, 10 and 20 years old were considered for comparison.
Pipes with 0 years old are considered to be as new pipe. For this pipe, only the
natural roughness of the pipe was considered during the analysis. Figure 4.41 shows
the values of the pressure variables for different ages of the pipe. Since the customer
requirement should have to be satisfied in terms of pressure in all groups of pipes,
the nodal pressures remain nearly the same for different ages of pipes. However,
slight increment in pressure drop is observed when the ages of the pipes increased.
159
Figure 4.41 Nodal pressures for different ages of pipes
The variation of flow variables with ages of pipes is shown in Figure 4.42. It is
observed that corrosion had a significant effect on flow capacity of the pipes. The
results of the simulation analysis showed that a decrease in flow capacity of 2.16%
and 4.35% were observed for the 10 and 20 years old pipes, respectively.
160
Figure 4.42 Flow through pipes for different ages of pipes
Table 4.10 shows the results of the simulation study conducted on pipe joining
node 0 and 1 of the pipeline network system. The friction factor, pipe flow resistance
and the flow through the pipe were studied under three age groups. As the age of the
pipes increased, the friction factor and pipe flow resistance increased. Based on flow
equations, for fixed pressure at the start and end node of the pipe, an increase in flow
resistance of the pipe reduced the flow capacity.
Table 4.10 Variation of flow with ages of pipes
Years in service
Friction Pipe flow resistance Flow[m3/hr]
New pipe 0.007003 9.01E-06 466683.00
10 0.007407 9.53E-06 456596.00
20 0.007847 1.01E-05 446371.00
161
4.4.3 Non-isothermal Flow Analysis
The analyses conducted in this section were based on the methodology discussed
in section 3.5.3 on page 101. The gunbarrel pipeline network shown in Figure 3.19
on page 77 was also analyzed with the aid of the enhanced Newton-Raphson based
TPNS simulation model for non-isothermal conditions. The same compressors data
as in section 4.3.1 was used for the analysis. The source temperature was assumed to
be 310 K. The results for the unknown pressure, flow and temperature variables were
obtained with less than ten iterations. The unknown temperature variables for the
first ten iterations are shown in Table 4.11. Note that temperature is measured in K.
Table 4.11 Results of temperature variables for the first 10 iterations
Iteration T1 T2 T3 T4 T5 Max. error 0 4000.00 4000.00 4000.00 4000.00 4000.00
1 360.576 466.115 97337.9 97443.5 98018.4 0.998674
2 309.982 237.188 237.263 -16406.9 -16538.2 0.602838
3 309.972 827.606 822.405 -40046.9 -39564.5 0.504891
4 309.961 581.863 575.621 13105.1 13051.8 0.310895
5 309.948 430.935 426.949 -2749.02 -2753.84 0.351659
6 309.939 362.365 360.53 926.084 917.115 0.107755
7 309.938 358.562 357 408.03 404.835 0.006933
8 309.938 358.469 356.91 412.797 409.561 2.79E-05
9 309.938 358.468 356.91 412.795 409.559 3.02E-09
10 309.938 358.468 356.91 412.795 409.559 1.52E-14
The convergence of the pressure variables and flow variables are shown in
Figure 4.43 and Figure 4.44, respectively.
162
Figure 4.43 Convergence of pressures for non-isothermal conditions
Figure 4.44 Convergence of flow variables under non-isothermal conditions
163
Based on the results obtained in Table 4.11, the convergence of some of the
nodal temperature variables for the network system is shown in Figure 4.45. The
convergences of the remaining nodal temperatures follow the same trends as T1. It
required a minimum of 6 iterations for the error to be stable. This is mainly due to
the initial estimations which were far from the final solution of temperature
variables.
Figure 4.45 Convergence of some of the temperature variables
4.5 Verification and Validation of the TPNS Simulation Model
The developed TPNS simulation model provided the required operational
variables of the pipeline network for various configurations. The results from the
simulation model needs simulation verification and validation for accuracy and
164
reliability. According to [94], model validation is usually defined to mean
“substantiation that a computerized model within its domain of applicability
possesses a satisfactory range of accuracy consistent with the intended application of
the model”. Model verification is often defined as “ensuring that the computer
program of the computerized model and its implementation are correct”. These
definitions of validation and verification are adopted in this thesis.
Since simulation models are increasingly being used in problem solving and in
decision making in various areas of interest, the verification and validation
techniques also varies depending on the types and nature of the model. The different
techniques on simulation model verification and validation are presented in
[95]-[97].
The availability of real system data gives more rooms to implement the various
verification and validation techniques on the simulation model. As presented in [98],
there are three situations concerning the data availability which includes, the case of
no data, the case of only output data, and the case of both input and output data
availability. Various model validations and verifications were suggested based on the
nature of the availability of the data.
The TPNS simulation model case is different and could not be categorized as the
situations reported in [98]. This is because, the TPNS simulation model is developed
based on the actual performance of the compressors, gas properties, pipe
information, and source and exit pressure requirements. As a result, some of the
input parameters are known. Even though the output data from the real system are
available, some data were considered to be confidential by the company. Hence, it
would be difficult to categorize the situations of the TPNS simulation model as one
of the classifications discussed based on the data availability.
Simulation model validation and verification are not considered as an end
process. They are integrated with the modeling process. Sargent [95] presented how
the simulation verification and validation process are integrated with the modeling
165
process. Sargent framework for modeling process is used as a basis in this research
work. Figure 4.46 shows the simplified framework for simulation modeling process.
By referring to Figure 4.46, the problem entity is the system (real or proposed),
idea, situation, policy, or phenomena to be modeled; the conceptual model is the
mathematical/logical/verbal representation (mimic) of the problem entity developed
for a particular study; and the computerized model is the conceptual model
implemented on a computer. The conceptual model is developed through an analysis
and modeling phase, the computerized model is developed through a computer
programming and implementation phase, and inferences about the problem entity are
obtained by conducting computer experiments on the computerized model in the
experimentation phase [95].
Figure 4.46 Simplified version of the modeling process [95]
4.5.1 TPNS Conceptual Model Validation and Verification
Conceptual model validation is one of the fundamental steps that were included
during the development of the TPNS simulation model. This phase of the model
development process includes the determination of the theories and assumptions
166
underlying the conceptual model are correct and that the model representation of the
problem entity is reasonable for the intended purpose of the model.
The various assumptions, theories and the relationships between parameters were
studied carefully in order to insure the conceptual model validity. Some of the
procedures conducted consist of thorough discussion with expert on the basic
principles for model development and studying the basic relationships between the
flow, pressure, temperature and performance measurements. Furthermore the whole
model was divided into sub-models and studied if appropriate structure, logic, and
mathematical relationships have been used. For instance, the pipe flow equations
which is one of the governing simulation equations was divided into various sub
groups in order to make sure that the mathematical relationships and basic principles
were properly implemented.
Proper examination of the flowchart of the model and plotting of the basic
relationships of the parameter were also conducted on the TPNS conceptual
simulation model. The general flow equation was plotted based on pressure versus
length of pipe for various flows in order to study the pressure drop profile. A
comparison with the known flow equations is also performed. The compressor
equation was derived by using some of the compressor data. The remaining data was
used to validate the compressor equations as shown in Figure 3.9.
The model verification procedures ensure that the computer programming and
implementation of the conceptual model are correct. There are various techniques
suggested in [95] for simulation model verification. Two types of tests were
conducted on the developed TPNS simulation model in order to ensure that the
computer program is properly implemented. The first test was done based on static
testing principles. This test consists of analyzing the computer program to determine
if it is correct by using correctness proofs, structured walk-through and examining
the structure properties of the program. The second test was conducted based on
dynamic testing principles. In dynamic testing, the computerized TPNS simulation
167
model was executed under different conditions like single phase flow analysis, two-
phase flow analysis, and single phase with no corrosion effect. Furthermore, the
TPNS model was tested for gunbarrel, branched and looped pipeline networks which
are the most commonly used configurations. The resulting values from the dynamic
analysis of the TPNS simulation model have shown that the computer program and
its implementations were correct. The techniques commonly used in dynamic testing
suggested in [94] are traces, investigations of input-output relations using different
validation techniques, internal consistency checks, and reprogramming critical
components to determine if the same results are obtained.
4.5.2 Operational Validation
One of the techniques applied for validation of TPNS simulation is the
operational validation techniques. This technique ensures whether the TPNS
simulation model’s output behavior has the required accuracy for the application of
the model to real system. This is where much of the validation testing and evaluation
take place. There are various simulation validation techniques suggested in
literatures [95], [97] and [99]. The type of validation techniques applied for the
simulation model depends on the nature of the simulation model and the availability
of data. The following sections discuss the detail of the operational validation
techniques applied on the TPNS simulation model.
4.5.2.1 Comparison to Other Models
One of the techniques used in order to validate the TPNS simulation model was
comparison with the previous models. The lack of data from literatures with the same
problem instances make difficult to validate the simulation model based on model
comparison. However, there were various types of pipeline networks analyzed for
determination of optimal parameters or determination of the nodal pressure and pipe
flow parameters for the given conditions. Two cases were taken from the available
literatures in order to validate the results of the developed TPNS simulation model.
168
As discussed in sections 4.3.1 and 4.3.3, the results of the TPNS simulation
model were compared with two other models based on various pipeline network
configurations. Even though complete data regarding the type of compressors used
were not mentioned in the cases considered, the TPNS simulation was able to
provide solutions in both instances. In the first case, the TPNS simulation model was
compared to an exhaustive optimization technique based on gunbarrel pipeline
networks system. As discussed in detail in section 4.3.1, TPNS simulation yielded
close solutions of nodal pressures and flow variables to the exhaustive technique.
The comparison based on looped pipeline network discussed in section 4.3.3 also
showed that the TPNS simulation model was able to provide solutions to nodal
pressures and flow variables which were close to previous model.
4.5.2.2 Validation based on Operational Graphics
As presented in [94], this type of validation techniques consists of showing
graphically the values of performance measures as the model moves through time or
subjected to variations. The dynamical behavior of performance indicators are
visually displayed as the simulation model runs with variations to ensure they behave
correctly.
The operational validation technique was demonstrated based on the pipeline
network shown in Figure 3.22. The performance of the compressor for the pipeline
network was presented in section 4.3.4. The performance analyses of the compressor
include the flow capacity, power consumption and compression ratio. The
performance characteristics maps generated using the developed simulation model
was similar to the one available in the literatures.
4.5.2.3 Model Parameter Variability - Sensitivity Analysis
One of the techniques used in order to validate the simulation model was
sensitivity analysis. This technique consists of changing the values of the input and
internal parameters of the model to determine the effect upon the model’s behavior
169
and its output. As presented in [100]-[102], performing sensitivity analysis on
simulation model is one of the key elements in studying the effect of the input
parameters variations on the output data.
Sensitivity analysis was conducted on the TPNS simulation model by varying the
input parameters like compressor speed, number of compressors working within the
compressor station and age of the pipe. The effect of the variation of these
parameters on the TPNS output parameters and model behavior was studied.
Sensitivity analysis was conducted on the part of the existing pipeline network
(Figure 3.22) which was used to demonstrate the application of the TPNS simulation
model for real system. The pipeline data shown in Table 4.7 was used for the
analysis. It was assumed that the demand pressure at various customer stations to be
4000kPa. The analysis was done based on single compressor data obtained from
[59]. The details of the sensitivity analysis conducted for the TPNS by varying the
input parameters are discussed as follows.
i) Variation in number of compressors working within the station
The variations of the number of compressors on the performance of the system
were studied based on constant speed operations of the compressors. Compressors
speed of 7500, 8000, 8250, and 8500rpm were taken as constant speed operations.
The maximum compression ratio of the system was limited to 1.5. Furthermore, the
source pressure for the pipeline network was 3000kPa and the age of the pipes was
assumed to be ten years.
Figure 4.47 shows the variation of main flow variable of the pipeline network
system (Q1) with number of compressors working within the station for different
speed operation of the compressors. When the number of compressors working
within the station is increased, it is observed that the flow capacity of the system
increased. As it is seen from equation (3.10), an increase in number of compressor
improves the flow capacity of the system. There will be more compressors to work
170
and satisfy the requirements. However, significant increment in flow rate capacity of
the system was observed when the number of compressors working within the station
increased to maximum of 4. The effect of the addition of a compressor to the existing
compressors became less as the number of compressors increased from 5. Therefore,
flow rate capacity of the system is more sensitive to increase in number of
compressors only at lower number of compressors.
Figure 4.47 Variations of flow rate with number of compressors
Figure 4.48 shows the compression ratios when the number of compressors
working within compressor station varies for different speed operations. An increase
in number of compressors working within the compressor station caused an increase
in compression ratio of the system for various speed operations. As seen from Figure
4.47, an increase in number of compressors enhanced the flow capacity Q1. Hence,
based on equation (3.1), an increase in flow capacity leads to higher pressure drop
which increase the CR. Significant changes in CR was observed for all speed
171
operations when the number of compressor changed from 1 to 2. Even though an
increase in number of compressors increased the compression ratio of the system,
their effect reduced as the number of compressors increased.
Figure 4.48 Variation of compression ratio with number of compressors
The magnitude of the variation of CR for various constant speed operations
became more sensitive to change in number of compressors. For instance, when the
number of compressor working within the system was 1, CR for compressor speed of
7500 rpm and 8500 rpm was 1.21459 and 1.24845, respectively. However, an
increase in number of compressors working within the system to 5 gave CR of
1.37902 and 1.50166 for compressor speed of 7500rpm and 8500rpm, respectively. It
is observed from Figure 4.48 that, compression ratio is more sensitive at lower
number of compressors and the deviations among the constant speed operations
increased with increase in number of compressors.
172
Figure 4.49 shows the variation of the power consumption with number of
compressors at various speed operations. In the case of the power consumptions with
number of compressors working, almost the same scenarios were observed as that of
the compression ratio variations.
Figure 4.49 Variation of power consumption with number of compressors
At lower number of compressors working within the station, power consumption
was more sensitive compared to high number of compressors working within the
system. The deviations in power consumption at various speed operations were
increased with increase in number of compressors working within the system.
ii) Variation in speed of the compressors working within the station
For performing the sensitivity analysis based on the variation in speed of the
compressors, the number of compressors working within the station was assumed to
be two. The source pressure for the pipeline network was 3500 kPa and the required
173
demand pressures were assumed to be 4000, 4250 and 4500kPa. The age of the pipes
was assumed to be ten years. The variation of flow capacity of the system,
compression ratio and power consumption with speed were studied and presented in
section 3.4.4.
iii) Variation in source pressure
For performing the sensitivity analysis based on the variation in source pressure,
three constant speeds operations of compressor were identified, i.e. 8000rpm,
8500rpm and 8750 rpm. The simulation can be performed any arbitrary speed
provided that the speed selected are within the working limit of the compressor. The
demand pressure for the system was 4500kPa. The age of the pipes was assumed to
be ten years. The number of compressors working within the compressor station was
two. The working range of source pressures for each speed is bounded by the
maximum CR and speed of the compressor. For instance, at source pressure of
3200kPa and compressor speed of 8750 rpm, the CR was 1.51074. Any source
pressures less than 3200kPa could result higher CR which is beyond the working
limit.
Figure 4.50 shows the variation of main flow rate (Q1) with variation in source
pressure (P0). For fixed compressor speed and demand pressure requirement, an
increase in source pressure increased Q1 and vice versa. For high speed operation of
the compressors, high increment in the main flow was observed compared to low
speed operations.
174
Figure 4.50 Variation of main flow rate with source pressure
As it is shown in Figure 4.51, an increase in source pressure decreased the
compression ratio of the system for constant speed operation and demand
requirement. In real system operation of the compressors, in order to meet fixed
demand pressure requirement at constant speed, an increase in source pressure makes
the system to move downward on the performance map of the compressor following
the trace of constant speed line. This causes an increase in flow rate capacity and a
decrease in compression ratio of the system. Thus, both the main flow Q1 and the
compression ratio of the system are highly sensitive to change in source pressure.
175
Figure 4.51 Variation of compression ratio with source pressure
Figure 4.52 shows the variation of power consumption of the pipeline network
system for various source pressures. An increase in source pressure increased the
power consumption the system. Power consumption by the system is mainly
dependent on the flow rate capacity of the system, compression ratio and properties
of the gas as presented in [23]. As shown in Figure 4.50 and Figure 4.51, an
increased in source pressure increased the flow capacity and decreased the
compression ratio of the system. Therefore, the overall effect of change in source
pressure is either to increase or decrease the power consumption of the system
depending on which factor is dominating. From Figure 4.52, it is observed that an
increase in source pressure increased the power consumption of the system which
indicates that the flow capacity of the system is the dominating factor over the
compression ratio.
176
Figure 4.52 Variation of power consumption with source pressure
iv) Variation in the age of the pipes
The variations of the flow capacity, compression ratio and power consumption of
the system with the age of the pipes were also studied. For performing the sensitivity
analysis based on the variation in the age of the pipes, the speed of the compressor,
source pressure, and the number of compressors working within the system were
assumed to be constant. The details of the effect of the age of the pipes on the
performance measures of the system were presented in section 4.4.2.
177
4.6 Summary
The simulation studies based on successive substitution and Newton-Raphson
method showed that the TPNS simulation model produced solutions to nodal
pressures and flow variables for various pipeline network configurations. Successive
substitution scheme was limited to simple pipeline network configurations as the
method is highly dependent on information flow diagram which is a difficult task to
produce when the number of unknowns increased. The applications of successive
substitution scheme on two different pipeline network configurations showed that, it
took many iterations to arrive at specified tolerance.
The simulation experiments were conducted on TPNS simulation model based on
Newton-Raphson solution scheme for the three most commonly found
configurations, i.e. gunbarrel, branched and looped pipeline network system. It took
less than ten iterations to arrive at reasonable error tolerance limit. Usually, it took a
maximum of 4 iterations for the TPNS simulation model for the error to arrive at the
stable solutions.
As discussed in sections 4.3.1 and 4.3.3, the results of the Newton-Raphson
based TPNS simulation model were compared with two other models based on
various pipeline network configurations. Even though complete data regarding the
type of compressors used did not mentioned in the cases considered, the Newton-
Raphson based TPNS simulation was able to provide solutions in both instances. In
the first case, the simulation model was compared to an exhaustive optimization
technique based on gunbarrel pipeline networks system. As discussed in detail in
section 4.3.1, simulation model yielded close solutions of nodal pressures and flow
variables to the exhaustive technique. The comparison of TPNS simulation model
based on looped pipeline network configuration was discussed in section 4.3.3. The
model produced solutions to nodal pressures and flow variables which were close to
previous model.
178
The application of the Newton-Raphson based TPNS simulation model for real
pipeline network system was also conducted based on existing pipeline network
system in sections 3.4.4. Three modules of TPNS simulation model which includes
input parameter analysis, function evaluation and network evaluation module were
evaluated using the data taken from the real system. Analyses of the performance of
compressor for existing pipeline network system which includes discharge pressure,
compression ratio and power consumption were also conducted using the developed
Newton-Raphson based TPNS simulation model. The performance characteristics
maps generated by the simulation model showed the variation of discharge pressure,
compression ratio, and power consumption with flow rate as similar to the one
available in the literatures.
The enhanced Newton-Raphson based TPNS simulation model was also applied
for analyzing branched and gunbarrel network configurations considering two-phase
gas-liquid flow analysis, single phase with corrosion effect and single phase with
temperature variations. In all cases, the results for the required nodal pressures,
temperature and flow variables were achieved with less than ten iterations to the
reasonable relative percentage errors. For instance, for two phase flow analysis
discussed in section 4.4.1 on page 144 for case 2C, the relative percentage error at
the end of the 10th iteration was obtained to be 8.16639x10-17.
As discussed in section 4.5, the TPNS simulation model was verified and
validated using the various techniques. The model verification and validation was
integrated with the TPNS modeling process. Conceptual model validation and
verification and operational validations were conducted on the TPNS simulation
model based on the available data and simulation experiment.
179
CHAPTER 5
CONCLUSIONS AND RECOMMENDATIONS
5.1 Conclusions
In order to evaluate the performance of natural gas pipeline transmission network
systems with non-pipe elements, a TPNS simulation model was developed. The
developed TPNS simulation model was able to provide solutions to nodal pressure
and flow variables for the given pipeline network systems. The performances of the
system were analyzed based on the variables obtained. The TPNS simulation model
was also enhanced to analyze pipeline network systems under two-phase gas-liquid
flow, variable ages of pipes, and variable temperature.
The developed TPNS simulation model incorporated detailed parameters of the
compressors into the governing simulation equations as discussed in section 3.2.3 on
page 52. Compressor stations should not be considered as a black box or represented
with few parameters during simulation. Speed of the compressor, flow rate, suction
pressure and discharge pressure were the critical elements which affected the
performance of the system as presented in section 3.4.4.
The TPNS simulation model proposed in this thesis based on the iterative
successive substitution and Newton-Raphson algorithm were tested for different
network configurations. The implementation of TPNS simulation model based on
successive substitution scheme discussed in section 4.2 showed that the model is
more suitable to simple network configurations. Simulation experiments were
conducted on TPNS simulation model based on Newton-Raphson solution scheme
for the three most commonly found configurations, i.e. gunbarrel, branched and
180
looped pipeline network system as presented in section 4.3. In all the investigations,
the solutions to the unknown variables were obtained with a wide range of initial
estimations. A maximum of 10 iterations were required to get solutions to nodal
pressure and flow variables with relative percentage errors less than 10-11.
The results of the Newton-Raphson based TPNS simulation model were
compared with two other models based on various pipeline network configurations.
In the first case, the simulation model was compared to an exhaustive optimization
technique based on gunbarrel pipeline networks system. The model yielded close
solutions of nodal pressures with less than 1.8% absolute parentage errors. The
comparison based on looped pipeline network also showed that, the Newton-
Raphson based TPNS simulation model was able to provide solutions to nodal
pressures and flow variables with mean absolute error of 5.10 % between the two
methods.
The application of the Newton-Raphson based TPNS simulation model for real
pipeline network system was also implemented based on the existing pipeline
network system as discussed in section 3.4.4. Three modules of TPNS simulation
model which includes input parameter analysis, function evaluation and network
evaluation module were evaluated using the data taken from the real system.
Analyses of the performance of compressor for existing pipeline network system
which included discharge pressure, compression ratio and power consumption were
also conducted using the developed TPNS simulation model. The performance
characteristics maps generated by the developed TPNS simulation model show the
variation of discharge pressure, compression ratio, and power consumption with flow
rate similar to the one available in the literatures.
Simulation analysis was conducted using the enhanced Newton-Raphson based
TPNS simulation model to investigate the effect of the ages of pipes on the
performance of the system based on gunbarrel and branched network configuration
as discussed in section 4.4.2. Pressure drop and flow rate of the gas were affected as
181
the age of the pipe increases. The comparison of the performances of the three
groups of pipes based on new, 10 years old, and 20 years old having branched
network configurations showed that a decrease in flow capacity of 2.16% and 4.35%
was observed for the 10 and 20 years old pipes, respectively. Since the customer
requirement have to be satisfied in terms of pressure in all groups of pipes, the nodal
pressures remain nearly the same for different ages of pipes as shown in Figure 4.41.
For instance the nodal pressure at node 1 increased by only 0.119% when the years
in service of the pipe increased to 10 years. The nodal pressure also increases by
0.239% when the years in service increased to 20 years.
The proposed enhanced Newton-Raphson based TPNS simulation model consists
of module to analyze pipeline network systems with two-phases by incorporating
modified homogeneous flow equation into the governing simulation equations. As
presented in section 4.4.1 on page 144 , the simulation analysis conducted on
branched TPNS configuration revealed that liquid holdup has a significant effect on
flow capacity of the system. For instance, for the main flow rate Q1, the flow
capacity reduced by 54.56% when the liquid holdup increased from 0.0001 to 0.005.
The corresponding nodal pressure at node 1 increased by 0.73%. Even though small
amount of liquid holdup, usually less than 0.005 [24], is found in transmission
pipeline network systems, the existence of the liquid had effect on pressure drop and
flow capacity of the system.
5.2 Contributions of the Research
The main contributions of the research are summarized as follows.
1. Performance analysis: The developed TPNS simulation model is able to
create alternative scenarios. There are two levels of creating alternative TPNS
scenarios. The first level of creating alternative scenarios involves varying
182
TPNS configurations. This includes varying the number of pipes involved,
number of compressor stations, the number of loops and the number of
junction points within the TPNS. The second level of generating alternative
networks involves varying the diameter of the pipes, the length of the pipes,
the number of compressors working within the stations, the type of
compressors, and the range of the speed of the compressors. The former
method is used to evaluate possible TPNS configurations in order to guide for
the selection of optimal network. The later method is used for evaluating the
operation of the existing system. Hence, the developed TPNS simulation
model could be used as tool for assisting decisions in designing and operating
TPNS.
2. Addressing the non-pipe elements: Investigation on various literatures on
TPNS simulation indicated that the overall operating cost of the system is
highly dependent upon the operating cost of the compressor stations in a
network [5, 6]. Hence, the detail incorporation of all its parameters, namely:
speed, suction pressure, discharge pressure, flow rates, and suction
temperatures are essential for a complete simulation of gas networks. The
proposed TPNS simulation model incorporated compressor stations by
integrating the characteristics map of compressor and energy equation as seen
in equation (3.10). Thus, this can provide more details for the analysis of
TPNS as presented in 4.3.4 on page 132.
3. Effect of age of the pipe: As discussed in section 4.4.2 on page 150, pressure
drop and flow rate of the gas are affected as the age of the pipe increases.
However, no known studies on the relationships between the age of the pipe
and its effect on pressure drop and flow were reported in the literatures. The
ability of the proposed enhanced Newton-Raphson based TPNS simulation
model to incorporate the age of the pipes in its flow equations as presented in
section 3.5.2 contributes for analyzing the effect of the age of pipes on
183
performance of the system. This could be useful for assisting decision in
maintenance and pipe replacement during the operation of TPNS.
4. Two phase flow analysis: When liquid exists in TPNS, single phase flow
modeling approach might not be adequate to predict the pressure drop, flow
capacity, and power consumption for the system. The proposed enhanced
Newton-Raphson based TPNS simulation model consists of module to
analyze pipeline network systems with two-phase by incorporating modified
homogeneous flow equation into the governing simulation equations as
presented in 3.5.1 on page 88. This could help to predict the transport
capabilities of the TPNS transporting two-phase gas-liquid mixtures.
5.3 Recommendations
The developed TPNS simulation model is mainly focused on natural gas
transmission pipeline network system. However, the principles used in TPNS
simulation model could be easily extended to be applied for the analysis of pipeline
network systems for other petroleum products with multiple sources.
The developed TPNS simulation model is able to make performance analysis for
the networks involving flow capacity, compressor ratio and power consumption for
the system. The addition of cost evaluation module to the current TPNS simulation
model could make the simulation model to analyze and evaluate the various network
configurations and operational scenarios based on the cost.
Pipeline network system mainly consists of pipes and many other non-pipe
devices such as compressor stations, valves and regulators. Although compressor
station is the key characteristics in the network, the TPNS simulation model takes
into account other non-pipe elements such as valves, regulators, and scrubbers could
be one area of research which needs further investigation.
184
In a pipeline network system problem, the system can be modeled as steady state
or transient model depending on how the gas flow changes with respect to time.
Transient analysis requires the use of partial differential equations to describe the
relationships between parameters. In case of transient simulation, variables of the
system, such as pressures and flows, are function of time. Transient TPNS simulation
model could be one of the problems to be addressed from the simulation perspective.
185
REFERENCES
[1] H. Ibrahim, A. S. b. Yaman, T. T. Lien, I. B. Santoso, and P. Laothumthut,
"Natural gas pipeline development in South East Asia", Asia Pacific Energy
Research Center, Tokyo, 2000.
[2] S. Wu, Steady-state simulation and fuel cost minimization of gas pipeline
networks, PhD Thesis, University of Houston, 1998.
[3] R. G. Carter, "Pipeline optimization: Dynamic programming after 30 years",
in Proceedings of the 30th PSIG Annual Meeting, Denver, 28-30 October,
1998.
[4] C. Borraz-S'anchez and R. Z. R'ıos-Mercado, "A Hybrid Meta-heuristic
approach for natural gas pipeline network optimization", LNCS, vol. 3636,
pp. 54-65, 2005.
[5] S. B. Kim, Minimum cost fuel consumption on natural gas transmission
network problem, PhD Thesis, Texas A & M University, 1999.
[6] R. Z. Ríos-Mercado, S. Kim, and E. A. Boyd, "Efficient operation of natural
gas transmission systems: A network-based heuristic for cyclic structures",
Computers & Operations Research, vol. 33, pp. 2323–2351, 2006.
[7] A. J. Osiadacz, Simulation and analysis of gas networks, first ed. Houston,
TX 77252, USA: Gulf Publishing Company, 1987.
[8] P. J. Wong and R. E. Larson, "Optimization of natural gas pipeline system
via dynamic programming", IEEE Transaction on Automatic Control, vol.
AC-13, No.5, pp. 475-481, 1968.
186
[9] M. Abbaspour, Simulation and optimization of non-isothermal, one
dimensional single/two phase flow in natural gas pipeline, PhD Thesis,
Kansas State University, 2005.
[10] K. F. Pratt and J. G. Wilson, "Optimization of the operation of gas
transmission system", Transaction of the instrument of the measurement
and control, vol. 6(5), pp. 261-269, 1984.
[11] M. M¨oller, Mixed integer models for the optimization of gas networks in
the stationary case, PhD Thesis, Technischen University Darmstadt, 2004.
[12] D. D. Wolf and Y. Smeers, "The gas transmission problem solved by an
extension of the simplex algorithm", Management Science, vol. 46, pp. 1454
-1465, 2000.
[13] J. J. Ruz, F. Saenz, and J. M. D. L. Cruz, "Modeling natural gas networks
for planning and scheduling", in 4th Workshop of Constraint Programming
for Decision and Control, Gliwice, Poland, 2002.
[14] T. v. d. Hoeven, "Constrained network simulation", in The 35th Annual
Meeting of Pipeline Simulation Interest Group (PSIG), Bern, Switzerland,
15 October - 17 October, 2003.
[15] K. C. Baumann, R. Alberchtsen, M. Carlsen, L. B. Rovang, and F. Romo,
"Transport analysis of the dry gas network system on the Norwegian
continental shelf using a network optimization tool", in The 37th Annual
Meeting of Pipeline Simulation Interest Group (PSIG), San Antonio, Texas,
7 November - 9 November, 2005.
[16] P. Nimmanonda, V. Uraikul, C. W. Chan, and P. Tontiwachwuthikul, "A
Computer-aided model for design of a simulation system for the natural gas
pipeline network system", in Proceedings of the 2002 IEEE Canadian
187
Conference on Electrical & Computer Engineering, Winnipeg, pp. 1634 -
1639, 2002.
[17] F. Abdolahi, A. Mesbah, R. B. Boozarjomehry, and W. Y. Svrcek, "The
effect of major parameters on simulation results of gas pipelines",
International Journal of Mechanical Sciences, 2007.
[18] P. Nimmanonda, A computer-aided simulation model for natural gas
pipeline network system operations, MSc Thesis, University of Regina,
2003.
[19] A. Valeriano and R. C. M. Castro, "Turbine and compressor performance
analysis through simulation", in 38th Annual Meeting of Pipeline Simulation
Interest Group (PSIG), Williamsburg, Virginia, 11 October -13 October,
2006.
[20] V. Patel, J. Feng, S. Dasgupta, P. Ramdoss, and J. Wu, "Application of
dynamic simulation in the design, operation, and troubleshooting of
compressor systems", in Proceedings of the thirty-sixth turbomachinary
symposium, pp. 95-105, 2007.
[21] F. W. Letniowski, "Compressor station modeling in networks", in 25th
Annual Meeting of Pipeline Simulation Interest Group (PSIG), Pittsburgh,
Pennsylvania, 1993.
[22] R. G. Worthingham, B. Asante, G. Carmichael, and T. Dunsmore, "Cost
justification for the use of internal coatings", in 25th Annual Meeting of
Pipeline simulation Interested Groups, Pittsburgh, Pennsylvania, 1993.
[23] E. S. Menon, Gas pipeline hydraulics, first ed. New York: CRC Press,
2005.
188
[24] B. Asante, Multiphase transport of gas and low loads of liquids in pipelines,
PhD Thesis, The University of Calgary, 2000.
[25] D. Fauer, "The making of useful simulation model", in The 34th PSIG
Annual Meeting, Portland, Oregon, 2002.
[26] B. Guo and A. Ghalambor, Natural gas engineering handbook, First ed.
Houston, Texas: Gulf Publishing Company, 2005.
[27] V. Chandra, Fundamentals of natural gas: An international perspective.
Tulsa, Oklahoma: PennWell Corporation, 2006.
[28] R. L. Busby, Natural gas in non technical language: Institute of gas
technology: PennWell Publishing Company, 1999.
[29] J. L. Kennedy, Oil and gas pipeline fundamentals, Second ed. Tulsa,
Oklahoma: PennWell Publishing Company, 1993.
[30] A. Lew and H. Mauch, Dynamic programming a computational tool:
Springer, 2007.
[31] R. Bellman, Dynamic programming, first ed. New York: Donver
Publications, Inc, 2003.
[32] P. J. Wong and R. E. Larson, "Optimization of tree-structured natural-gas
transmission networks", Journal of Mathematical Analysis and Application,
vol. 24, pp. 613-626, 1968.
[33] R. Z. Rios-Mercado, S. Kim, and E. A. Boyd, "Optimizing fuel
consumption on gas transmission networks: Evaluation of a heuristic for
cyclic systems", IEEE, pp. 3118-3122, 2001.
189
[34] P. B. Percell and M. J. Ryan, "Steady-state optimization of gas pipeline
network operation", in The 19th Annual Meeting of Pipeline Simulation
Interested Group (PSIG), Tulsa, pp. 684-690, 1987.
[35] A. J. Osiadacz and M. Gorecki, "Optimization of pipe sizes for distribution
gas network design", in The 27th Annual Meeting of Pipeline Simulation
Interest Group (PSIG), Albuquerque, New Mexico, 1995.
[36] Y. Wu, K. K. Lai, and Y. Liu, "Deterministic global optimization approach
to steady state distribution gas pipeline networks", Springer
Science+Business Media, vol. (8) 259 -275, 2007.
[37] W. S. Chmilar, "Statistical modeling techniques in the design and operation
of pipeline systems", in 34th Annual Meeting of Pipeline Simulation Interest
Group (PSIG), Portland, Oregon, 2002.
[38] C. K. Sun, C. W. Chan, and P. Tontwiachwuthikul, "A fuzzy expert system
for optimizing pipeline operation", IEEE, pp. 777-780, 1997.
[39] C. K. Sun, V. Uraikul, C. W. Chan, and P. Tontiwachwuthikul, "An
integrated expert system/operations research approach for the optimization
of natural gas pipeline operations", Engineering Applications of Artificial
Intelligence, vol. 13, pp. 465-475, 2000.
[40] V. Uraikul, C. W. Chan, and P. Tontiwachwuthikul, "Development of an
expert system for optimizing natural gas pipeline operations", Expert system
with applications, vol. 18, pp. 271 -282, 2000.
[41] V. Uraikul, C. Chan, and P. Tontiwachwuthikul, "Hybrid intelligent system
development: A case study for decision support of natural gas pipeline
operation", in Proceedings of the 1999 IEEE Canadian Conference on
Electrical and Computer Engineering, Edmenton, Alberta, 1999.
190
[42] A. Chebouba, F. Yalaoui, L. Amodeo, A. Smati, and A. Tairi, "New method
to minimize fuel consumption of gas pipeline using ant colony optimization
algorithms", in proceedings of IEEE conference on Service System and
Service Management, pp. 947-952, 2006.
[43] T. Mora and M. Ulieru, "Minimization of energy use in pipeline operations-
an application to natural gas transmission systems-", in proceedings of the
31st annual conference of IEEE, pp. 2190-2197, 2005.
[44] P. D. Surry, N. J.Radcliffe, and I. D. Boyd, "A Multi -objective approach to
constrained optimization of gas supply networks: The COMOGA method",
Springer-Verlag LNCS, vol. 993, pp. 166-180, 1995.
[45] J. Stoffregen, K. K. Botros, D. J. Sennhauser, K. Jungowski, and H.
Golshan, "Pipeline network optimization-application of genetic algorithm
methodologies", in 37th Annual Meeting of Pipeline Simulation Interest
Group (PSIG), San Antonio, Texas, 7- 9 Nevember 2005.
[46] A. J. Osiadacz, "Dynamic optimization of high pressure gas networks using
hierarchical systems theory", in 26th Annual Meeting of Pipeline Simulation
Interest Group (PSIG), San Diego, California, 1994.
[47] S. Wu, R. Z. Rios-Mercado, E. A. Boyd, and L. R. Scott, "Model
relaxations for the fuel cost minimization of steady-state gas pipeline
networks", Mathematical and Computer Modeling, vol. 31, pp. 197-220,
2000.
[48] R. Z. Rios-Mercado, S. Wu, L. R. Scott, and E. A. Boyd, "A Reduction
technique for natural gas transmission network optimization problems",
Annals of Operations Research, vol. 117, pp. 217-234, 2002.
[49] J. Mohring, J. Hoffmann, T. Halfmann, A. Zemitis, G. Basso, and P.
Lagoni, "Automated model reduction of complex gas pipeline networks", in
191
36th Annual Meeting of Pipeline Simulation Interest Group (PSIG) Palm
Springs, California, 2004.
[50] C. A. Chung, Simulation Modeling Hand Book: A Practical Approach. New
York: CRC PRESS, 2004.
[51] H. P. Bloch, Compressors and modern process applications. Hoboken, New
Jersey.: John Wiley & Sons, Inc., 2006.
[52] P. C. Hanlon, Compressor Handbook, First ed. New York: McGraw-Hill
Companies, 2001.
[53] M. Bryant, "Complex compressor station modeling", in 29th Annual Meeting
of Pipeline Simulation Interest Group (PSIG) Tuscon, Arizona, 1997.
[54] S. Wright, M. Somani, and c. Ditzel, "Compressor station optimization", in
30th Annual Meeting of Pipeline Simulation Interest Group (PSIG), Denver,
Colorado, 1998.
[55] S. P. d. Santos and M. Lubomirsky, "Gas Composition Effect on Centrifugal
Compressor Performance", in 35th Annual Meeting of Pipeline Simulation
Interest Group (PSIG), Williamsburg, Virginia, 2006.
[56] R. Brown and K. Rahman, "Turbine/Compressor performance monitoring
software and flow capacity", in 34th Annual Meeting of Pipeline Simulation
Interest Group (PSIG), Portland, Oregon, 2002.
[57] R. G. Carter, "Compressor station optimization: computational accuracy and
speed", in 28th Annual Meeting of Pipeline Simulation Interest Group
(PSIG), San Francisco, California, 1996.
[58] R. G. Carter, D. W.Schroeder, and T. D. Harbick, "Some causes and effects
of discontinuites in modeling and optimizing gas transmission networks", in
192
The 25th Annual meeting of Pipeline Simulaiton Interes Group (PSIG),
Pittsburgh, Pennsylvania, October 1993.
[59] R. Kurz and S. Ohanian, "Modeling turbomachinery in pipeline
simulations", in 35th Annual Meeting of Pipeline Simulation Interest Group
(PSIG), Bern, Switzerland, 2003.
[60] M. T. Gresh, Compressor performance Aerodynamics for the user, second
ed. Boston: Butterworth Heintemann, 2000.
[61] B. Asante, "Two phase flow: accounting for the presence of liquids in gas
pipeline simulation", in 34th Annual Meeting of Pipeline Simulation Interest
Group (PSIG), Portland, Oregon, 23 - 25 October, 2002.
[62] B. Asante, "Multiphase transport of gas and low load liquids in pipelines",
in 32th Annual Meeting of Pipeline Simulation Interest Group(PSIG),
Savannah, Georgia, 2000.
[63] I. R. Ellul, G. Saether, and M. E. Shippen, "The modeling of multiphase
systems under steady state and transient conditions-a tutorial", in 36th
Annual Meeting of Pipeline Simulation Interest Group (PSIG), Palm
Springs, California, 20 October - 22 October, 2004.
[64] T. S. Golczynski, "Defining operating envelopes for multiphase pipelines-a
flow assurance approach ", in 35th Annual Meeting of Pipeline Simulation
Interest Group (PSIG), Bern, Switzerland, 13 October - 15 October 2003.
[65] N. H. Boriyantoro and M. A. Adewumi, "An integrated single phase/two
phase hydrodynamic model for predicting the fluid flow behaviour of gas
condensate in pipelines", in 26th Annual Meeting of Pipeline Simulation
Interested Group (PSIG), San Diego, California, 13 - 14 October, 1994.
193
[66] M. Leksono, One-dimensional compositional modeling of gas and
condensate flow in pipelines, PhD Thesis, The Pennsylvania State
University, 1990.
[67] G. Stanley and V. Vanderford, "Application of on-line simulation for a two
phase wet gas pipeline in Indonesia", in 32th Annual Meeting of Pipeline
Simulation Interest Group (PSIG), Savannah, Georgia, 2000.
[68] Y. Taitel and A. E. Dukler, "A model for predicting flow regime transitions
in horizontal and near horizontal gas-liquid flow", AIChE, vol. 22, No.1 pp.
47- 54, 1976.
[69] R. J. Wilkens, Prediction of the flow regime transitions in high pressure,
large diameter, inclined multiphase pipelines, PhD Thesis, Ohio University,
1997.
[70] O. Shoham, Mechanistic modeling of gas-liquid two-phase flow in pipes.
222 Palisades Creek Drive Richardson, TX 75080-2040 USA: Society of
Petroleum Engineers, 2006.
[71] B. Azzopardi and J. Hills, Modeling and experimentation in two-phase flow
vol. 450. New York: Springer Wien 2003.
[72] A. J. Osiadacz and M. Chaczykowski, "Comparison of isothermal and non-
isothermal pipeline gas flow models", Chemical Engineering Journal, vol.
81, pp. 41 -51, 2000.
[73] J. Modisette, "Pipeline thermal models", in 34th Annual Meeting of Pipeline
Simulation Interest Group (PSIG), Portland, Oregon, 2002.
[74] R. W. Revie and H. H. Uhlig, Corrosion and corrosion control: an
introduction to corrosion science and engineering, Fourth ed. Hoboken
New Jersey: John Wiley & Sons, Inc., 2008.
194
[75] P. A. Ferrira, M. F. C. Bezerra, and R. Loschiavo, "The internal corrosion
integrity strategy on the development of new offshore production areas in
Brazil", in Offshore Technology Conference Houston, Texas, U.S.A, 2004.
[76] D. P. Riemer, Modeling cathodic protection for pipeline networks, PhD
Thesis, University of Florida, 2000.
[77] X. Zhou, Experimental study of corrosion rate and slug flow characteristics
in horizontal, multiphase pipeline, MSc Thesis, Ohio University, 1993.
[78] S. Nestic, S. Wang, J. Cai, and Y. Xiao, "Integrated CO2 corrosion-
multiphase flow model", in Corrosion 2004, Houston, 2004.
[79] P. Ghosh, Numerical methods with computer programs in C++, first ed.
New Delhi: Prentice-Hall, 2006.
[80] S. Salleh, A. Y. Zomaya, and S. A. Bakar, Computing for numerical
methods using visual C++, first ed.: John Wiley & Sons, 2008.
[81] W. F. Stoecker, Design of thermal systems, third ed.: McGraw-Hill
International Editions, 1989.
[82] V. R. Subramanian, B. S.Haran, and R. E.White, "Series solution for
boundary value problems using a symbolic successive substituion method",
Computers & Chemical Engineering, vol. 23, pp. 287-296, 1999.
[83] M. Abbaspour, K. S. Chapman, and L. A. Glasgow, "Transient modeling of
non-isothermal, dispersed two-phase flow in natural gas pipelines", Applied
Mathematical Modelling, vol. 34, pp. 495-507, 2010.
[84] D. Brkic, "An improvement of Hardy Cross method applied on looped
spatial natural gas distribution networks", Applied Energy, vol. 86, pp.
1290-1300, 2009.
195
[85] S. B. M. Beck and F. R, Boucher, "Steady state analysis of fluid circuits
containing three port devices", Trans IChemE, vol. 75, pp. 73-80, 1997.
[86] M. Kessal, "Simplified numerical simulation of transients in gas networks",
Trans IChemE, vol. 78, pp. 925-931, 2000.
[87] A. D. Woldeyohannes and M. A. A. Majid, " Development of Simulation
Model for Natural gas Transmission System", in International Conference
on Plant Equipment and Reliability, Kuala Lumpur, 27-28 March 2008,
2008.
[88] M. A. A. Majid and A. D. Woldeyohannes, "Simulation model for the
analysis of transmission pipeline network system of natural gas", in 2nd
Engineering Conference on Sustainable Engineering Infrastructures
Development & Management, Sarawak , Malaysia, 18-19 Dec 2008, 2008.
[89] "As Tested Performance curves for Petronas GPP 5 & 6 Dresser -Rand
Model 755B7 Sales gas compressor No. K5 -0501A ": Petronas Gas Berhad,
1996.
[90] A. D. Woldeyohannes and M. A. A. Majid, "Simulation model for
optimization of pipeline network system for natural gas transmission", in
International conference on Science &Technology, Pinang, Malaysia, 12-13
Dec 2008, 2008.
[91] A. D. Woldeyohannes and M. A. A. Majid, "Simulation model for natural
gas transmission pipeline network systems", Simulation Modeling Practice
and Theory, SIMPAT-D-09-222, Submitted on June 2009 (under review).
[92] A. D. Woldeyohannes and M. A. A. Majid, "Simulation of Natural Gas
Transmission Network System Performance", Journal of Energy and Power
Engineering, vol. 3, No.12 (Serial No.25), Dec. 2009.
196
[93] R. Kurz, "The Physics of Centrifugal Compressor Performance", in The
34th Annual Meeting of Pipeline Simulation Interest Group (PSIG), Palm
Springs, California, 20 October - 22 October, 2004.
[94] R. G. Sargent, "Verification and validation of simulation models", in
Proceedings of the 1998 Winter Simulation Conference, pp. 121-130, 1998.
[95] R. G. Sargent, "Verification and validation of simulation models", in
Proceedings of the 2008 Winter Simulation Conference, pp. 157-169, 2008.
[96] J. P. C. Kleijnen, "Theory and methodology: verification and validation of
simulation models", European Journal of Operational Research, vol. 82,
pp. 145-162, 1995.
[97] P. L. Knepell and D. C. Arangno, Simulation validation: a confidence
assessment methodology, First ed.: IEEE Computer Society Press, 1993.
[98] J. P. C. Kleijnen, "Validation of models: statistical techniques and data
availability", in Proceedings of the 1999 Winter Simulation Conference, pp.
647-653, 1999.
[99] A. M. Law and M. G. McComas, "How to build valid and credible
simulation models", in Proceedings of the 2001 Winter Simulation
Conference, pp. 22-29, 2001.
[100] K. Chan, A. Saltelli, and S. Tarantola, "Sensitivity analysis of model output:
Variance based methods make the difference", in Proceedings of the 1997
Winter Simulation Conference, pp. 261-268, 1997.
197
[101] J.-M. Furbringer and C.-A. Roulet, "Confidence of simulation results: put a
sensitivity analysis module in your model The IEA-ECBCS Annex 23
experience of model evaluation", Energy and Buildings, vol. 30, pp. 61-77,
1999.
[102] J. P. C. Kleijnen, "Case study: statistical validation of simulation models",
European Journal of Operational Research, vol. 87, pp. 21-34, 1995.
RESULTS OF TPNS SIMULATION MODEL FOR GUNBARREL NETWORK
*** W E L C O M E T O T P N S S I M U L A T I O N ***** This Program Analyses a TPNS with 3 pipes and two CSs serving single customer** *** P I P E L I N E D E T A I L S ** Input the Length of the pipes [km] 80 80 80 Input the diameter of the pipes[mm] 900 900 900 Input the service life of the pipes [year] 0 0 0 *** P I P E L I N E F R I C T I O N F A C T O R ** 0.00700345 0.00700345 0.00700345 *** P I P E L I N E F L O W C O N S T A N T S ** 7.21113e‐006 7.21113e‐006 7.21113e‐006 *** C O M P R E S S O R S T A T I O N D E T A I L S ** Input the number of the compressors within the station 1: 1 Input the number of the compressors within the station 2:1 Input speeds of the compressors 8000 8000 *** C U S T O M E R S P E C I F I CA T I O N S ** Input demand pressure requirements at customer each station PD1 =4000 Enter the initial estimates for all unknowns: **Note that the first 4 elements are pressure parameters and the next element is flow parameters** Enter the initial estimation for the unknown parameters:4000 Enter the number of iterations to go:10 THE SOLUTION IN 2 iteration(s) x_new[0] = 3103.27 x_new[1] = 3573.37 x_new[2] = 3551.63 x_new[3] = 4021.73 x_new[4] = 3.01569e+006 The tolerance in 2 iterations =0.998674 THE SOLUTION IN 3 iteration(s) x_new[0] = 383.496 x_new[1] = 3994.87 x_new[2] = 1709.82 x_new[3] = 6020.34 x_new[4] = 1.88147e+006 The tolerance in 3 iterations =0.602838 THE SOLUTION IN 4 iteration(s) x_new[0] = 976.174 x_new[1] = 3902.12 x_new[2] = 2849.15 x_new[3] = 5036.5 x_new[4] = 1.25024e+006
The tolerance in 4 iterations =0.504891THE SOLUTION IN 5 iteration(s) x_new[0] = 2062.99 x_new[1] = 4313.44 x_new[2] = 3620.2 x_new[3] = 4694.88 x_new[4] = 953728 The tolerance in 5 iterations =0.310895.THE SOLUTION IN 6 iteration(s) x_new[0] = 2450.26 x_new[1] = 3550.52 x_new[2] = 3036.26 x_new[3] = 4386.49 x_new[4] = 705598 The tolerance in 6 iterations =0.351659. THE SOLUTION IN 7 iteration(s) x_new[0] = 2471.58 x_new[1] = 3505.17 x_new[2] = 3064.83 x_new[3] = 4346.65 x_new[4] = 636963 The tolerance in 7 iterations =0.107755. THE SOLUTION IN 8 iteration(s) x_new[0] = 2472.77 x_new[1] = 3505.01 x_new[2] = 3065.89 x_new[3] = 4345.74 x_new[4] = 632577 The tolerance in 8 iterations =0.00693339. THE SOLUTION IN 9 iteration(s) x_new[0] = 2472.77 x_new[1] = 3505.01 x_new[2] = 3065.9 x_new[3] = 4345.73 x_new[4] = 632559 The tolerance in 9 iterations =2.78791e‐005 x_new[0] = 2472.77 x_new[1] = 3505.01 x_new[2] = 3065.9 x_new[3] = 4345.73 x_new[4] = 632559 The tolerance in 10 iterations =4.48679e‐010 THE SOLUTION IN 11 iteration(s) x_new[0] = 2472.77 x_new[1] = 3505.01 x_new[2] = 3065.9 x_new[3] = 4345.73 x_new[4] = 632559 The tolerance in 11 iterations =1.02997e‐015 The solution within the specified number of iteration is : x_new[0] = 2472.77 x_new[1] = 3505.01 x_new[2] = 3065.9 x_new[3] = 4345.73 x_new[4] = 632559 The compression ratio for the first station is = 1.41744 The compression ratio for the second station is = 1.41744 The energy consumption in KW for the system is = 15855.6 : Press any key to continue
SUCCESSIVE SUBSTITUTION BASED SIMULATION RESULTS (CASE 1A)
Intr. No P1 P2 P3 Q1 QC2 QC1 0 - - 6000.00 100.00 - - 1 3678.87 6000.04 3666.79 6291.69 12899.49 -12799.49 2 2434.75 3910.86 1814.21 12715.58 7199.71 -908.02 3 2216.57 3293.23 3538.23 13340.78 510.76 12204.81 4 3106.64 4564.45 2419.12 10095.38 6865.21 6475.57 5 2107.34 3257.87 2415.34 13622.00 3642.51 6452.87 6 2605.59 3808.31 3075.67 12159.87 3629.74 9992.26 7 2697.67 4045.72 2429.77 11833.15 5620.64 6539.23 8 2339.99 3527.85 2717.10 12998.04 3678.31 8154.84 9 2643.73 3908.45 2765.30 12026.99 4587.10 8410.94 10 2525.27 3795.38 2560.44 12428.92 4731.16 7295.84 11 2485.71 3711.21 2749.06 12556.30 4103.91 8325.01 12 2593.01 3863.06 2664.90 12202.98 4682.82 7873.48 13 2501.94 3749.52 2646.67 12504.45 4428.83 7774.15 14 2536.94 3782.87 2712.73 12390.72 4372.96 8131.49 15 2551.62 3812.07 2654.48 12342.23 4573.96 7816.76 16 2517.00 3763.40 2678.15 12455.83 4396.93 7945.30 17 2544.66 3797.50 2685.79 12365.29 4469.23 7986.60 18 2535.14 3789.06 2664.78 12396.63 4492.46 7872.83 19 2529.74 3779.00 2682.37 12414.33 4428.47 7968.16 20 2540.62 3794.13 2675.73 12378.60 4482.09 7932.24 21 2532.30 3783.98 2672.87 12405.94 4461.89 7916.71 22 2534.91 3786.13 2679.53 12397.39 4453.15 7952.79 23 2536.82 3789.53 2674.20 12391.11 4473.44 7923.95 24 2533.39 3784.81 2676.04 12402.37 4457.22 7933.89 25 2535.89 3787.83 2677.08 12394.17 4462.81 7939.56 26 2535.20 3787.31 2674.98 12396.44 4466.00 7928.16 27 2534.53 3786.17 2676.58 12398.63 4459.59 7936.85 28 2535.61 3787.64 2676.08 12395.10 4464.48 7934.15 29 2534.86 3786.75 2675.71 12397.55 4462.96 7932.14 30 2535.04 3786.86 2676.38 12396.96 4461.83 7935.73 31 2535.27 3787.24 2675.89 12396.21 4463.85 7933.12 32 2534.93 3786.78 2676.03 12397.32 4462.38 7933.84 33 2535.15 3787.05 2676.16 12396.59 4462.78 7934.54 34 2535.11 3787.03 2675.95 12396.73 4463.18 7933.41 35 2535.03 3786.90 2676.09 12396.99 4462.54 7934.19
36 2535.14 3787.04 2676.06 12396.64 4462.98 7934.00 37 2535.07 3786.96 2676.01 12396.86 4462.88 7933.77 38 2535.08 3786.97 2676.08 12396.83 4462.74 7934.12 39 2535.11 3787.01 2676.04 12396.74 4462.94 7933.88 40 2535.07 3786.96 2676.04 12396.85 4462.81 7933.93 41 2535.09 3786.99 2676.06 12396.78 4462.84 7934.01 42 2535.09 3786.99 2676.04 12396.79 4462.88 7933.90 43 2535.08 3786.97 2676.05 12396.82 4462.82 7933.97 44 2535.09 3786.99 2676.05 12396.79 4462.86 7933.96 45 2535.09 3786.98 2676.04 12396.80 4462.85 7933.93 46 2535.09 3786.98 2676.05 12396.80 4462.84 7933.97 47 2535.09 3786.98 2676.05 12396.79 4462.86 7933.95 48 2535.09 3786.98 2676.05 12396.80 4462.85 7933.95 49 2535.09 3786.98 2676.05 12396.80 4462.85 7933.96 50 2535.09 3786.98 2676.05 12396.80 4462.85 7933.95